Stack of grammar books

Do you really understand your customers?

stack of language and grammar booksPoets, philosophers and scientists have spent generations exploring the question of what makes humans, human.

Laughter? Love? The inclination to text and drive?

For the artistic and scientific, the basic question provides the inspiration for works of great beauty and deep analysis. But today, the question, in a slightly more crass form, has been taken up by marketers — marketers who are not as much interested in the universal answer as they are in the individual answer.

Marketers in the age of digital transformation want to know what makes that particular human, human. And more importantly what makes that particular human tick? It is a question at the root of developing real-time, one-to-one personalization.

In my search for answers, I bypassed the poets and philosophers. I politely sidestepped the marketers and turned to the scientists, in particular a data scientist. Amit Aggarwal is BloomReach’s chief technology officer and a guy who has thought a lot about personalization and its relationship with human intent and human nature.

Aggarwal naturally has a deep technical understanding of the algorithms and data streams that make authentic personalization a real thing. But he indulged me, taking a broader view of personalization and breaking it down into its vital components:

  • Semantic language understanding
  • Contextual understanding
  • User behavior understanding
  • Product understanding

Aggarwal had compelling arguments for each. Me? I was fixated on coming up with the most important element of personalization. Aggarwal wasn’t biting. The world isn’t simple. Multi-faceted problems aren’t solved with one solution. Complicated challenges aren’t met with simple solutions. A data scientist knows this.

And the greatest of these is language

A writer does not. And so among the fearsome foursome of personalization, I decided the greatest of these is language. Imagine, a writer choosing language as the key to success. But hear me out. Nothing is more human that language. It sets us apart from the other animals.

It is the basis for all we do. Grand ideas are nothing without a way to express them. Love goes unnoticed if it’s not expressed. The next big thing falls silently in the forest if no one can describe it.

And so, a compromise: I’ll respect Aggarwal’s superior understanding of personalization by presenting all four basic elements of true personalization. But I’ll start with my favorite — language.

First, however, why this whole “what is personalization” exercise in the first place? Because personalization, as a term, needs some refinement. It has become a buzz phrase to mean whatever the one uttering it wants it to mean.

I look at personalization as the ability to understand a consumer on an individual basis; the ability to understand a consumer’s intent in the moment, and the corollary ability to provide that consumer with relevant and personal content as a result.

And that brings us back to Aggarwal’s four elements, starting, naturally, with language.

Semantic language understanding: True personalization begins with a deep understanding of what digital users are saying. One of the deepest human needs is to be understood. And that understanding is achieved through language. Not just words, but language. Language comes with sentiment and context. Language changes. “That’s sick,” meant one thing in 1950. It meant something entirely different in 2000.

The future of personalization

  • Join us at BloomReach Connect in New York to hear customer experience visionaries talk about personalization in the age of digital transformation.
  • Connect banner

There is no personalization without language. Language is different from words. Plenty of technologies can understand words. You type in a search query and the machine looks for product descriptions, for instance, that contain those words.

Language is more than that. Language is living and ever-changing. Different digital consumers use words differently. In a sense we all speak our own language. Not only does a real personalization system need to acknowledge that fact, it needs to constantly learn to better understand what language means and what consumers mean by the language they use.

Language includes synonyms, different words that mean the same or similar things. True personalization understands that the user searching for a “camel pocketbook” and the user searching for a “tan purse” might be interested in the same product or the same story about a fashion trend.

Contextual understanding: But when a user says or types “tan purse,” how does a personalization system understand whether the user is searching for a product or a fashion story or even instructions on how to make a tan purse or advice on what colors go best with a tan purse? The system needs to understand context. It needs to understand what a digital user hopes to accomplish.

Is she seeking inspiration? I she researching a product or subject after being inspired? Is she already sure of a product she wants to buy and hunting for the seller with the best price or fastest or most convenient delivery options? Has she already bought a tan purse and is looking for care instructions or whether she can return the purse for a different color or style?

True personalization understands context and constantly learns from interactions with millions of users. It understands that someone in Northern California searching for “water shoes” in the summertime is not likely to be looking for footwear designed for a rainy day. Instead, it’s quite likely that person is searching for footwear meant for the beach and for wading into the water.

User behavior understanding: One part of grasping context is understanding using behavior and what a user’s digital activity and the activity of users across the web tells us about their intent. Here, true personalization parallels the work of a high-quality store associate.

When you walk into a home-improvement store, the human associates who work there can learn a lot about you. Yes, they learn from what you say to them. But they can begin building their understanding before a customer even opens his or her mouth. What aisles and products do you pause in front of? Do you browse one aisle and then head to another with products that might be used in the same DIY project?

Do you examine the top-of-the-line variable speed drill and then put it back and reach for a serviceable model that costs less? Do you stride through the store with a certain confidence? Or do you look a little lost, eyeing the 1-inch, 90-degree PVC elbow segment like it’s something an Apollo crew brought back from the moon?

A perceptive associate, which granted is sometimes hard to find, knows a lot about you already. Or at least he or she knows a lot about you and your intent on this particular shopping trip.

A constantly learning machine that deeply understands language can draw some similar conclusions from online behavior. Moreover, understanding not just what words mean, but how they relate to other words and what their connotations are in different circumstances, means that an artificial-intelligence-driven system can serve up recommendations that are relevant to one, single individual.

Product understanding: Deeply understanding a user’s language and behavior and the context in the moment is a start to providing true personalization. But it isn’t enough. A true personalization system needs to understand what a user is in the market for. Is someone who’s shown an interest in financial products looking for a home loan or a credit card? Is the user who’s shown an intent to learn more about insurance, looking for life insurance or homeowner’s insurance?

A true personalization system must deeply understand content in order to properly match a consumer’s intent with the available content. There is no deep user understanding without a deep understanding of the content they are seeking.

Consumers are the ones defining personalization

What’s crystal clear is that in today’s world, it’s consumers who are deciding what “personalization” means.

And so it’s only right that enterprises listen to consumers and provide what it is their customers are demanding. Not only is it the right thing to do, it’s also the profitable thing to do. It’s clear that the ability to understand language will only become more important for enterprises as the era of digital transformation rages on.

Total Retail Report found that commerce sites that based their site search on keyword search, returning only recommendations that included the keywords, suffered a 40 percent abandonment rate. Sites that rely on semantic understanding experienced only a 2 percent abandonment rate, Total Retail Report said.

A few other statistics help explain why digital consumers can become so frustrated. Google has said that 15 percent of the search queries it receives on any given day are queries that the search engine has never seen before. It’s reasonable to imagine that individual sites are subject to the same sort of effect, meaning the site needs to be smart enough to understand the meaning of what a searcher is saying.

For in the end, we are individuals who celebrate our own individuality in what we believe, how we dress, whom we associate with, where we work and yes, how we talk. In English alone there are more than 1,000 ways to say something is beautiful and more than 1,500 ways to say “happy.”

No doubt we will continue to find new ways to express ourselves, whether we’re talking about beauty, happiness or buying a new pair of shoes. It will be up to those building digital experiences to keep that in mind as they design better ways to keep up.

Photo of books by Jackie Finn-Irwin published under Creative Commons license. 

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.


How to find the right digital tools for your business, from KPMG’s omni-expert


As the person who leads consultancy KPMG’s global omni-business practice, Duncan Avis knows in excruciating depth and detail the challenges businesses in all sorts of industries face.

The world is moving faster and faster, with no sign of letting up. The demand for results and the need for profits remain relentlessly constant. And the speed with which competitors can zig and zag accelerates by the quarter.

And that’s before we even get to the issue of customers, which are the reason a business is in business, or should be. Let’s just say, they’re not making things any easier.

“This is a difficult world today,” Avis says, “where customers are constantly changing their expectations.”

I was lucky enough to get a chance to talk with Avis at the National Retail Federation’s annual Big Show earlier this year. He shared his thoughts on what it means to be an “omni-business,” which in fact requires a greater degree of difficulty than being omni-channel, a staple of digital retailers’ strategies and conversations.

NRF being a massive gathering of retailers, I asked Avis if he could break down the biggest challenge retailers face in 2017. You’re forgiven if even just hearing his answer makes you a little tired.

“I think one of the challenges is that there are a lot of challenges,” Avis said. He elaborated in the video clip below.

One way for retailers and others to begin to address those challenges, Avis told me, is to expand the way they think about “omnichannel.”

“We think about it as ‘omni-business,’” he says. “And the reason why we call it that is because we really think it’s about connecting the front of the office, middle, and back office, of any enterprise.”

Doing business right, it seems, goes beyond simply connecting with customers. It also means connecting the moving parts of an enterprise in a way that promotes a widely held vision and understanding of how to deliver on the promises the company is built on.

As it is, Avis says, retailers and consumer goods companies tend to say they place omnichannel or omni-business as a high priority. But not even a third say they are getting the expected return on investment from their efforts.

Again, that is most likely because their omni efforts are focused on the front office and not on the entire organization. All of which made the case in my mind for the embracing the idea of a chief digital officer, an executive responsible for making sure customers receive a memorable digital experience — memorable in a good way.

But Avis argues that the idea of a chief digital officer is too limiting. Digital should be an imperative throughout the organization, as he explains in the video clip below.

Given that we were at NRF, surrounded by 235,000 square feet of exhibitors, many of them pushing technology solutions for selling, marketing and retaining customers, I asked Avis if he had any advice for those who need to sift through the offerings and decide what technology to buy and deploy.

In the video below, he says that retailers shouldn’t start with the technology. Instead, they should think about the experience they want to provide consumers.

All of which raised another question in my mind: Among the technologies discussed and displayed at NRF were robots that would help in-store shoppers find what they were looking for and Internet of Things advances that would automate ordering and purchasing.

And while they all were certainly cool in a gee-whiz sort of way, I wondered whether retailers, many of whom are facing existential pressures, were in any sort of position to sprint to the cutting edge while the core of their businesses were so vulnerable.

The answer, no doubt, is that it depends. It depends on the particular retailer, its needs, its customers’ needs and the like. Avis added that it’s important to define some of the terms that are thrown around. For instance, exactly what are people talking about when they’re talking about “robots?”

Learn more about success through Human+Machine

“Robotics can mean so many different things,” he says. “Where we think there’s a tremendous opportunity, is the use of artificial intelligence and robotic process automation to actually streamline processes, to use it to better understand consumer behavior, to really get beyond what the known knowns are, to get to the unknown unknowns. But that doesn’t necessarily need to be a physical robot. There is tremendous technology in the sense of AI that can really enable an employee, a human employee, to be much more effective and successful in their relationship with a customer.”

The big opportunity with artificial intelligence, Avis says, is to use it to augment human performance — the powerful combination of human and machine.

When it comes to the deep and difficult challenges that retailers face, artificial intelligence offers one key way forward — a key way that will no doubt be combined with others as digital retailers and businesses move forward at a pace far faster than yesterday and much slower than tomorrow.

Photo of Duncan Avis by Mike Cassidy

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Pini Yakuel, Optimove

Video: As digital retail transforms, so do those doing the work

Retailers of all sizes are scrambling to keep up with rapid shifts in technology and in customer habits and expectations.

Walk around a retail trade show, like the annual NRF Big Show, (OK, we did) and you’ll be bombarded with the phrase “retail transformation.” Everyone knows the industry is changing at a maniacal pace. The big question, of course, is what should retailers do about it?

We caught up with two experts at NRF, both of whom made the point that retailers, and digital retailers in particular, have to find new ways to think about their jobs and how they organize their enterprises.

Optimove founder and CEO Pini Yakuel said retailers need to do better at getting past their fixation on acquiring customers. Sure, attracting consumers is important, but if you are acquiring a customer who makes one purchase then disappears, you’re not doing your business any favors. In the video below, Yakuel talks about the “gold mine” of data that retailers have concerning existing customers and how best to use it, so you don’t need to acquire the same customers over and over again.

For his part, Dan Chester, ForeSee’s vice president of retail sales, says digital retailers are grappling with the best ways to keep up with customer expectations. Chief marketing officers, for instance, are juggling a mind-boggling number of technologies and all the various constituencies that rely on them.

Their big challenge, he says in the video below is making sure all the pieces and people are integrated and working toward the same goals.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Pini Yakuel, Optimove

Video: Artificial intelligence helps build relationships

Optimove founder and CEO Pini Yakuel says artificial intelligence — despite its cold, number-crunching reputation — can actually help build emotional bonds between businesses and their customers.

We caught up with Yakuel, an Optimove co-founder, at the National Retail Federation’s Big Show in January, where he spoke with us about how amplifying human effort with machines can help marketers better understand their customers and thereby become more responsive.

This is the third video in our ongoing series. See more BloomReach NRF coverage on the blog.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Collision: Contemplating the singularity and discussing suicide with a robot

Silicon Valley veteran Jerry Kaplan came to the Collision tech conference in New Orleans to talk about artificial intelligence, human intelligence, the likelihood of singularity and the future of technology and our lives.

Jerry Kaplan

But his message was a little more tangible than all that: Don’t let the hype around artificial intelligence distract you from the tremendous opportunities and potential pitfalls that lie before us as human beings. We are a long way away from humans merging with machines or machines taking over human existence.

“We need to get rid of, in this field, this gee-whiz, apocalyptic gloss,” Kaplan, an artificial intelligence pioneer who co-founded Go and Onsale, said speaking from the conference’s main stage. “I’m not worried about super-intelligent machines or whether I’m going to live long enough to be uploaded into cyberspace. Such concerns only distract us from the real threats and opportunities that mankind is likely to face.”

More Collision coverage

It’s the sort of discussion that breaks out at Collision, a three-year-old conference that attracts entrepreneurs, investors, media and is increasingly becoming a place for the technerati to be seen.

While the vast exhibit hall is full of startups showing off how they intend to change an industry, a part of an industry or the world, the conference sessions include presentations not only about where tech is taking us all, but also whether that is a place we want to be.

Kaplan talked about two world’s of artificial intelligence, or AI: one very grounded, powering business worldwide and evolving at times in different directions. The other one of near science fiction in which machines “are going to get so smart that they’re going to stop at nothing to achieve their goals.”

He argued for the former view. Yes, progress is being made, but artificial intelligence progress can be measured in refinements and improvements, not some steady march to free-thinking machines.

Arguing that fully thinking machines are one step closer every time artificial intelligence takes a step forward is a little like saying your smartphone is becoming more intelligent every time you download an app, Kaplan argued. (Hey, now it knows the weather outside! Hey now it can tell my sports scores!)

None of which diminishes the achievements of those developing artificial intelligence — which now drives cars, offers shoppers personalized recommendations, writes financial earnings stories, shows you the way home, recommends restaurants and on and on, Kaplan said. Nor does it diminish the importance of artificial intelligence in our future.

“I think it’s an important technology,” he said, “and it’s going to have a very big impact, as big an impact, in my view, as the invention of the wheel.” Kaplan’s talk was one of several on Wednesday that reminded technologists that with great power comes great responsibility.

Natalie Monbiot, of UM Worldwide, opened a separate panel on the commercial applications of artificial intelligence by in effect laying out the stakes.

“We’re basically entering — AI is helping us to enter — the holy grail of marketing, where we can actually have, as brands, emotional one-to-one relationships with consumers at scale.”

el kaliouby and Monboit

Great for marketers, right? What’s better than appealing not only to a customer’s mind, but to a customer’s mind and heart. But what comes with that power?

Rana el Kaliouby co-founded Affectiva, a company that provides a tool to recognize emotion by a person’s voice and facial expressions. The technology allows companies to craft experiences that acknowledge and incorporate those emotions.

“Emotions are very personal to people,” she said. “If  if we start building these very emotionally engaging experiences, what kind of responsibility does that place on the designer of the experiences? How do you incorporate empathy?”

The panel talked about a future of robotic healthcare-givers in homes and robots taking care of children and chatbots that will gather medical information from patients. What sort of requirement is there for transparency? What do those being served by machines need to know about how those machines operate and what they do with information?

Monbiot suggested that some studies indicated that patients are more comfortable “talking” to machines about uncomfortable subjects, like depression and suicide.

Is that good news? el Kaliouby said it’s good that patients can open up, but then what?

“Now the avatar may know you have depression,” she said. “Does it disclose this to your doctor? Does it tell your mom? Does it tell your partner?”

Fascinating questions, no doubt. And in some ways, not unlike the sorts of ethical and legal dilemmas that arise almost whenever technology moves forward. In fact, the technology often moves faster than the mores that will govern it.

It’s a positive, then, that the conversation is underway, even as the work on artificial intelligence continues on. No doubt, the future will be here before we know it — whether it comes with the singularity or not.

Photo of Jerry Kaplan courtesy of Collision. Photo of Rana el Kaliouby, Babak Hodjat, Douglas Merrill and Natalie Monbiot by Mike Cassidy.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Let’s hear it for the humans at Collision tech conference

At the Collision conference, an annual explosion of tech talk, future chatter, investor wrangling and start-up flogging, it’s easy to feel as though you are lost in the machine.

Machines are everywhere, if not physically then narratively in talks about robot ethics, the computational future, and “DIY VR,” which is “do-it-yourself virtually reality,” which, when you think about it, sounds a lot like reality reality.

Screen Shot 2016-04-26 at 4.05.00 PM

And no doubt, the machines are key to the continuing digital revolution. The machines are synonymous with the future, artificial intelligence, the Internet of Things, smart homes, smart cars, smartphones, natural language processing and data-driven decisions.

Paddy Cosgrave, the founder of Collision, which is holding its third U.S. gathering in New Orleans this week, even warmly embraced the machine in a blog post, which explained how data-driven the hyper-socializing and networking is at Collision. In fact, even the lanyards that the 11,000 expected this year are wearing were born of data-driven design, he wrote.

More Collision coverage

But if you listen closely, you will hear the sound of human voices and see the results of human effort. The world, it turns out, is not black or white, 0s or 1s, humans or machines. The world works best when it turns to human+machines — systems that take the best of both and combine them into a powerful force.

The idea took center stage, literally, during a Collision conversation between Brandon Berger, of marketing giant Ogilvy and Wired magazine’s Spencer Reiss. Berger had tons of examples of how the machines (and data more specifically) had upended the world of advertising and marketing again and again. And I’ll get to some.

But then Reiss gave him a little nudge and asked about storytelling. (“When you’re trying to do the touchy feely emotional marketing things, how the hell do you scale it?” is how he put it. With that, Berger agreed that creating the stories that become the ads or the promotions or multi-faceted campaigns really need the human touch.

Yes, he explained an agency like Ogilvy leans on technology to determine what platforms to use, to understand where to reach audiences, to determine what resonates with them. But someone has to plant the seed, to have the idea, to create the content.

“I mean, creating the lightning in a bottle of a brilliant idea that goes viral, we are much more in tune with what’s going to work now with data,” Berger said. “But that’s not going to happen all the time, and I think, for us, extending a message and telling a great story on many platforms, it really is a lot of work in terms of creating and refining that story.”

Brandon Berger and Spencer Reiss

Implicit in Berger’s comments is the existence of the mixture of art and science that all kinds of enterprises — including advertisers, retailers, sports teams — have embraced.

+Aziz Ali, a New Orleans musician and marketing strategist, speaking at Collision later this week, told me that many musicians feel the internal struggle of following their muse while following their metrics.

“Reality is, that for musicians, we’re living in an increasingly DIY-oriented world,” said Ali, a 32-year-old guitar player. “And what you sort of realize over time is, especially as your career kind of moves forward, is that the creative artistic endeavors, sitting with your instrument, the performing, that part of it, it shrinks. So much of your time ends up being, figuring out what you’re doing on different social platforms, assessing why you should be there and more importantly, how you’re doing on there and how you should be measuring yourself.”

But it’s part of the gig now, he says, if you want to meet your fans where they are. (Read more about Ali’s music+marketing life in this post.)

Musicians, of course, are hardly alone in recognizing the evolution from a time when humans drove business decisions to a time when the machines are picking up more of the workload. Let’s pick back up with Berger talking about marketing and advertising campaigns.

“Back in the day of ‘Mad Men,’ there were planners and strategists who took qualitative insights and came up with a theory of how those things would move markets and sell products,” Berger said.

That’s changed — a lot.

“Now what is so amazing is that technology and data is so accessible,” he added, “that we can look at insights like search, or human behavior or social behaviors and take those ideas and, in real time, create content to then help our brands infiltrate a market.”

Berger’s proof point: diaper cakes. No, seriously. Diaper cakes. Ogilvy, which apparently represents a diaper company, found that people by the thousands were using the search term “diaper cake,” which are cakes made of diapers, meant to be given as a baby shower gift, not eaten. Anyway, it was instant campaign fodder.

But still, it took people to figure out what to do with this diaper cake information — and presumably to have the good sense to know not only that such a cake was inedible, but also highly unappetizing.

So, despite the rise of the machines and all the good that’s done us, humans will remain a vital part of the equation and a significant part of the conversation at Collision.

Certainly, as long as we have diaper cakes.

Photo of Reiss and Berger by Mike Cassidy. Robot illustration from BloomReach archive.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.


Big data gets personal at Structure Data 2016

If the technological revolution has taught us anything, it’s to expect more, more, more.

And faster.

I’m here to tell you, that as demanding as we have all become, we are not going to be disappointed. My optimism springs from spending a day last week at the Structure Data 2016 conference in San Francisco. There, some of the companies that have most intimately intertwined themselves with our everyday lives talked about what they’re cooking up to make money, while making our lives easier.

It’s not so much what I heard in individual presentations that has stoked my optimism. A lot of what I heard were visions that have been floating around for awhile — and the 20-minutes-per-session format made it difficult to take a deep dive into most of the ideas.

Instead, it was the fact that I was immersed in a day full of conversations featuring Facebook, Yahoo, Uber, Netflix, Pandora, Pinterest, Slack, Microsoft and the Orlando Magic all talking about what they are doing and where we are headed.

Yes, the Orlando Magic — and I can explain.

But first, think about how much the world has changed in the last two decades. The companies I listed include enterprises that make our everyday tools, provide our diversions; our everyday pleasures and, occasionally, frustrations. They’ve become for us what CBS, Sears Roebuck, Life magazine, Bell Telephone, Smith-Corona, Metro-Goldwyn Mayer, Checker Cab and a host of other brand names were for previous generations.

And think about how much the world will change in the decades to come. If my day at Structure Data is any indication, we haven’t seen anything yet. These companies, primarily tech companies, are pushing forward at a mad pace, exploring the myriad ways that harnessing data can change our lives — presumably for the better.

If I were to grab onto a theme of the day, I’d pick personalization, because in some form or another, many of the coming breakthroughs discussed at Structure Data have to do with making information, services and products more relevant to us as individuals.

Some examples from among many:

Pandora and Gracenote

old radio

Few would argue with the notion that Pandora has a firm grip on personalization. Chris Martin, Pandora’s chief technical officer, threatened to explode a few heads by opening his presentation with the fun facts that the digital radio station has a database of 65 billion pieces of feedback (think thumbs up/thumbs down) from listeners and 1.4 million human-curated music tracks that help train machines to curate even more tracks.

Oddly, one of Pandora’s personalization projects is to come up with a way to personalize for a group of people, if, in fact, that is personalizing. Ty Roberts, of Gracenote, who shared the stage with Martin, explained the problem, which his company is also focused on.

“If I’m in the car, I’m listening to Black Sabbath,” he said. “But if my wife gets in the car, there is no Black Sabbath. So what I listen to in the medium range is the Eagles.”

But let’s say the couple then picks up Roberts’ 16-year-old son. He’s more a Kendrick Lamar guy. The Eagles are a no-go.

“That’s still a problem,” Roberts says. “So, Gracenote is trying to figure that out. How to map all that together.”

Gracenote maintains a huge database of information about the contents of compact discs, which allows digital devices to recognize and identify musical tracks. It also sells systems to automatically create playlists and recommend music.

“I think most consumer media companies are having this problem,” Martin said of appealing to multiple users simultaneously. “At the moment, profile management is hard.”

Think of Netflix, he said. All it takes to mess up his profile is to have his kid watch one movie under his name, he said.

And so, Pandora is looking at a way to combine personalization and context. The opportunity, Martin said, is for Pandora to gather more data about where and with whom listeners are listening to music and to come up with ways to understand and acknowledge that.

Gracenote is focused on a similar initiative, Roberts said.

“It’s going to take into account the weather,” he said. “You’ll get a playlist that will relate to what you’re doing in the car. What are you listening to on your way to work? What are you listening to on the way home?”


newspaper reader

Yahoo news is in business to personalize your news feed. But recommending relevant news stories is a tricky business, Suju Rajan, Yahoo’s director of research for personalization science, told moderator Signe Brewster. What’s relevant in the morning before work, might not be relevant on the way home — and not just because news by its nature is perishable.

The truth is that people on the way to work are in a hurry. The content they consume might be very specific to what they need to do or know that day. On the way home — and once home — they might have time to slow down a bit, to read deeper into a subject and explore new topics.

Someone reading on a laptop might be willing to click on a story to quickly check it out. A mobile user might want to read only headlines and keep scrolling without clicking at all.

Like the Pandora problem, it’s all about context.

“We’re trying to figure out when you want to engage in them,” Rajan says of stories and videos, “as opposed to when you want to really be in the know. I think that would be a cool thing for us to tackle.”

And how does Yahoo begin to tackle that? By training machines that then train themselves based on users’ behavior. But even doing that requires thoughtfulness. What user behavior is important?

“While we were building the personalized news feed, one thing we understood early on is that click behavior is hardly an indication of user interest,” Rajan said. “The amount of time that a user spends reading an article is what is indicative of their interest.”

I’ve seen the problem described as the “click web” vs. the “attention web.”

As a result, Yahoo pays a lot of attention to users who come back to the site frequently in, say, a three-month period — long-term engagement, they call it. What do such site visitors have in common?

“If we optimize for dwell-time, the amount of time a user spends, and rank them that way, that is highly correlated with the fact that the user will keep coming back to the stream, over and over again,” Rajan says.



For its part, Facebook is working on better understanding the ways individuals talk. The idea is to make the social network more conversational. Alan Packer, who goes by director of engineering, language tech at Facebook, told interviewer Stacey Higginbotham that Facebook’s machines are very good at understanding text. But spoken language is more difficult.

People have accents, speak different languages, use idioms, swear, use slang, shorthand and made-up words.

But those conversational styles are going to be more and more important to understand, because users are moving away from sitting down at a keyboard to compose posts or post pictures etc. They are moving to mobile devices and even devices with no keyboards at all, Packer said. (Apple co-founder Steve Wozniak talked about the trend, using the Amazon Echo as an example.)

And so, spoken language is becoming a key to communicating on the social network.

Think of the sort of written conversations going on on Facebook today.

“Dialogs are taking place all the time on Facebook,” Packer said. “‘Hey, I’m going to Las Vegas next week. What’s the hot restaurant?’ Where we’re heading toward, is Facebook actually wants to participate in that conversation and share helpful information.”

And more importantly, advertisers would like to share helpful information — information that would help you choose their restaurant for dinner, for instance. But soon those conversations will be more likely to be spoken and Facebook needs to deeply understand them.

Maybe, like Higginbotham, the idea that Facebook wants to join your conversations doesn’t sound like pure upside to you. Packer assured Higginbotham that users would remain in control of the conversation.

“You can totally unfriend us,” he said.

Orlando Magic

Alex Martins Orlando Magic

The NBA’s Magic this season has moved big time into big data to set ticket prices, create a more memorable customer experience and build loyalty.

Magic CEO Alex Martins ran through the steps the team is taking with John Paul of VenueNext, which is overseeing the team’s efforts to use data to build a better in-stadium experience.   

I know what you’re thinking: Isn’t putting a winning team on the court the best way for an NBA franchise to create a memorable experience and build loyalty? But if putting together a winning team were easy, everybody would be the Golden State Warriors. And the Magic are not the Golden State Warriors. In fact, they are at the bottom of their division, having lost more games than they’ve won.

But besides playing basketball, the Magic also sells stuff — food, drink, team gear, parking and game tickets. So, like others in the retail world, they’ve stepped up their game when it comes to customer experience.

“We’re transforming what used to be a season-ticket purchase into a membership,” the Magic’s Martins told the Structure Data crowd.

For a set price ($499 for the rest of this season), a fan can become a Magic member and attend every home game. The catch: The fan doesn’t know exactly where his or her seat is until he or she walks into the arena and is sent their seat assignment, which is determined by supply and demand. If the Magic are playing a popular opponent and the crowd is big, the member will end up in the cheap seats. If not?

“If it’s a low-demand game,” Martins told me after his presentation, “you’re probably going to end up with a great seat.”

At the core of the Magic’s data transformation is a team app with features that the Magic started rolling out about a year ago. Season ticket holders can use the app to turn unused game tickets into credits that can be spent at concession stands and on tickets for future games. The app can also be used to buy tickets, upgrade tickets, order food and pay for parking.

“We are collecting an enormous amount of data now on all the people who are using the app,” Martins said. “We can tailor a customer’s experience to each individual, based on the data that we’ve gathered on the previous purchasing experience.”

Here’s a video of the Martins’ session:

Paul, of VenueNext, said 80 percent of the Magic’s season-ticket holders use the app and that 26 percent of Magic tickets are sold through the app. He said the app is six times more popular than a sports team app typically is.

Martins said downloads of the new app are up 300 percent over the old version and that 75 percent of fans have used the app to purchase something — food, beverages or retail items — while in the arena.

“Food and beverage sales are up 15 percent,” Martins told me. “That’s totally attributable to the people who are using the app to order.”

The app and in-arena sales are part of a larger data strategy that weighs supply and demand in setting the price for game tickets.

“Our ticket prices, based on demand and based on our data, change by the minute,” Martins added. “It’s based upon the demand of the opponent, the day of the week, certainly how our team is playing, etc.”

Team employees also keep an eye on the behavior of season ticket holders — are they turning tickets in; are they buying food and drinks while at the games or do they appear to be losing interest?

“We created a road map for our service reps to figure out which of our clients they needed to spend the most time with to hopefully convert them to renewing for next season.”

And about next season… Martins assured those gathered at Structure Data that the team was in fact also using data to try to take its players’ game up a notch. Like other NBA teams, the Magic crunch data to track players’ fitness, analyze defensive sets, help players identify the best spots on the floor from which to shoot and the best places to position themselves on the court to snag rebounds.

But that, frankly, is a data discussion for another day. Maybe next year.

Photo of radio by Robert Couse-Baker and newspaper reader by Connie Ma published under Creative Commons license. Photo of Facebook sign courtesy of Facebook. Photo of Alex Martins with John Paul and Tom Krazit by Mike Cassidy.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

GE shows how data has tranformed business

Quick: What do you think of when you think of GE, the company formerly known as General Electric?

Huge grey machines that grind and spark and hum, producing electricity? Dirty diesel-powered locomotives rumbling and screeching through the countryside pulling miles of coal cars?

old school locomotive

Me too. But really what GE is, it turns out, is an ideal prism through which to look at the future. We all know the world is changing at warp speed. But sometimes the change is so rapid, so vast and so seemingly diffuse, that it helps to occasionally zero in on one story to see the bigger picture.

Which brings us to William Ruh, GE’s chief digital officer, speaking on stage with the New York Times’ Steve Lohr at the Structure Data 2016 conference in San Francisco.

“Everything we’re doing is on the bleeding edge of technology,” Ruh said, talking about how the digital revolution has revolutionized GE. “It’s a cornerstone of our culture. We’re not an industrial company. We’re a digital industrial company.”

So, put away your images of workers shoveling coal into GE-built boilers or whatever those hulking grey machines in your mind’s eye actually are. GE is much more into the types of machines that artificial intelligence experts talk about — the learning machines, the kinds of machines that fly airplanes, run the economy, power digital marketing and e-commerce, inform medicine and on-and-on.


Granted, my conference epiphany is based on a 20-minute talk by one executive with one company. But the kinds of change that Ruh talked about reflected a much broader change. For instance, those big machines you picture (OK, I picture) when you think of GE?

That’s not really the way GE looks at the world, in part because that’s not how GE’s customers, and others who buy things, look at the world.

“We’re going to see a movement away from big capex projects,” Ruh said, using the shorthand for “capital expenditure.”

Those who run companies, by and large, no longer see products when they buy something. Instead, they buy outcomes, he said. An electricity provider, for instance, no longer buys a power plant. And an airplane manufacturer no longer buys a jet engine.

“They think about the idea, ‘I can generate 5 percent more electricity on a wind turbine or I can get more hours out of my jet engine,” Ruh said. “This digital thing is fundamentally changing our company.”

Of course, GE still sells stuff. But like so many legacy companies before it, it increasingly sells services and expertise. It’s selling outcomes, Ruh said, an idea that “is fundamentally embedded into how you build machines and how  you service them. This idea of selling outcomes; how far will that go?”

The question was rhetorical, but the answer is: Pretty far.

It is the digital revolution and the ability to gather, process and analyze huge amounts of data that have changed the nature of GE’s business and business in general.

wind turbines

Ruh talked about a wind farm that GE worked with that was initially producing 5 percent in additional electricity. By using data-fueled computer modeling to optimize the production of the existing turbines, GE boosted that additional electricity to 20 percent — and nearly doubled the power provider’s profit, he said.

How? GE takes real-time data — accounting for weather and the performance of the turbine blades, for instance — applies sophisticated algorithms and physics modeling to create multiple digital “twins” of each turbine in a wind field. It then tests a range of operating conditions and practices and instantly comes up with the most efficient way to produce power.

The company can use the same method, Ruh said, to create digital twins of every aspect of a power plant to improve efficiency throughout the plant. And, he says, there is no reason to think such digital twins will extend only to the industrial world.

“I contend that in 30 years we’ll have a digital twin of a human and all this data we collect, that we really can’t do much with, we will feed into it,” he said of the digital human model.

Optimizing the human body. Pretty cool.

So just how far has GE evolved? At the end of Ruh’s talk, Lohr diplomatically asked about recruiting tech talent to a company that might be seen by some (ahem) as an old-school, industrial company. How would he sell GE to, say, a Silicon Valley tech whiz deep in the land of Google, Facebook, Apple and the many startups that turning the world upside down?

“You get to work on meaningful things,” said Ruh, who left Silicon Valley giant Cisco to work for GE.

And yes, these days, the line works.

Photos of wind turbine by TLPOSCHARSKY and steam engine by David Lofink published under Creative Commons license. Conference photo of William Ruh and Steve Lohr by Mike Cassidy.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.


Video: How should retailers use big data?

In our third installment of BloomReach University, Video Campus, Brian Beck, senior vice president of e-commerce and omni-channel strategy for Guidance, talks about the role of data in e-commerce and retail.

Beck shares some thoughts about how to handle data silos and offers a reminder that customers are not all that concerned with what is happening behind the scenes. They just want to buy products.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Video: IBM talks about cognitive computing and the future of retail

Welcome to the inaugural class of BloomReach University, Video Campus. Over the next several weeks, we’ll be posting short videos (hey, you can watch them in line at the grocery store), featuring thought leaders in retail, e-commerce and digital marketing, who will provide their takes on key trends and challenges in the industries they follow.

No need to take notes. (You can just replay the video.) But there will be a quiz. OK, there won’t.

First up is Patricia Waldron, director of global retail solutions for IBM, talking about the era of cognitive computing and how that’s changing retail. She’ll also touch on ways in which the ability to gather, analyze and quickly act on data — on clicks, on foot traffic and on how consumers react to marketing appeals — is changing the way the game is played.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

IBM Watson tries his hand at retail

IBM Watson is learning the same lesson that so many American workers have learned in the modern era: In the course of a lifetime, you are likely to have multiple careers.

Watson, of course, has had storied careers as a “Jeopardy” champion and a medical researcher (including veterinary medicine apparently). And now the cognitive computing system is setting its sights on a retail career.

The big lug, who’s palled around with Bob Dylan in T.V ads is one of the stars at the National Retail Federation’s Big Show 2016, which opened today in New York. One of the early sessions featured Watson’s work with The North Face, which in December rolled out a feature that let’s online customers shop for coats by carrying on a typed conversation with Watson, the way they might converse with a sales person when they go into a store.

In a way Watson is an appropriate celebrity face for this week’s trade show. He represents two of the key characteristics of retail now and into the future. The first is the need to provide one-to-one personalization for shoppers. The second is the wisdom in combing  humans and machines to tackle the personalization problem at large scale.

(More on Watson’s place in personalization.)

That said, being a celebrity isn’t always easy. In his work with The North Face, Watson encountered a problem common to many famous folks: His reputation preceded him.

More coverage of NRF’s Big Show 2016

“The system requires a lot of teaching,” Cal Bouchard, The North Face’s senior director of e-commerce told the crowd during a Sunday morning session on Watson. “As an AI (artificial intelligence) newbie, I thought, ‘Oh, Jeopardy champion. Watson. I’m going to get that. I’m going to get this thing that beat Ken Jennings and has all the information in the whole world already built in. Not true.”

It is going to take time for Watson to do all the things Bouchard wants him too, but she said she’s taking the long view. After all, the best solutions often aren’t the easiest solutions. For now, the Watson feature works only for coat shopping and it is better at asking questions (and modifying results based on answers) than it is at answering them.

I used Watson to shop for a coat for a trip to Chicago in February. It’s top recommendation was a spiffy model in the color I told Watson I liked, but it cost $399. When I asked Watson if he had anything cheaper, like a good commission sales person he said: “I don’t understand what you mean by ‘cheaper.’”  

IBM Watson

But so it is with developing technology and developing uses for it. Bouchard is willing to wait. (She also knows something about celebrity and expectations, having been co-captain of the 2000 Canadian Olympic basketball team.)

“We’re always looking for ways to engage the consumer that may not be the quickest hit to purchase or conversion, but maybe a long-term play,” she said.

And she sees a day when Watson will not only be making product recommendations, but also providing ski tips for those buying ski jackets and offering travel suggestions for those who mention that they’re buying a coat for a trip.

“We’re just at the beginning of looking at how AI technology can really apply to retail,” added Neil Patil, of digital agency Fluid, who presented Sunday with Bouchard. “ We know that AI technology is going to be huge. And it’s huge that we consider that in the retail industry.”

Watson exhibit photo by Edward Blake published under Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.


Collision conclusion: Big data for the masses is near

Sometimes at big, chaotic tech conference it’s best to look for inspiration in unlikely places — like the title of one of the dozens of sessions meant to draw from among the mob scurrying from demo to meeting to to talk to funding pitch.

And so it was at the Collision conference in Las Vegas this week, which offered a session called, “It’s the Data, Stupid.” No, it’s hardly an epiphany that data is a big deal. But that wasn’t the point of the session nor of the many conversations I had with attendees about the state of data today.


What’s clear is that we are nearing a tipping point with big data, a new era in which the tools, technology and practices exist to democratize data. More and more frequently, the insights that analyzed data provide can be used by teams throughout an organization to make immediate decisions that help move a business forward.

Listen to Amanda Kahlow, founder and CEO of 6Sense, talk about the new ways businesses view the potential of data.

“Five years ago, people were investing in the back end technology,” Kahlow, whose company provides predictive analytics for marketers, said during a Collision panel that I moderated Wednesday. At the time, technologies like Hadoop were all the rage and received all the buzz, she said.

“Today,” Kahlow continued, “it’s all about the applications. How do we use this data to drive something forward? And how do we use it got get measurable results?”

Yes, in Jeopardy-like fashion, she phrased her answer in the form of a question. And it’s true that the democratization of data has a ways to go. The vision of just the right insight being available to just the right person at just the right time is a work in progress. But plenty of tools are out there and they are being put to good and profitable use.

Take the example of marketers, as Richard Frankel, president of Rocket Fuel, a programatic media-buying company, did during the aforementioned “It’s the Data, Stupid” session.

“For marketers, the job is changing,” he said. “It used to be a ton of people guessing and doing research: of the torrent of data, which tiny bits could they use? At human scale, that was all the data they would spend time with. Now, with all the technology that’s come along, all the assumptions of how much data could be used and how fast it could be used have gone out the window.”

Machine learning, natural language processing, the ability to wrangle unstructured data and render it meaningful has changed everything.

It’s as if data is moving through an evolution that is familiar in the world of technology. When the first personal computers were developed. Visionaries were wowed. A complete computer that you could fit on a desk! For others, the question was: But what can you do with it? Then came spreadsheets and word processors, which helped democratize the PC. And then the Internet, which broke everything wide open.



Hundreds of companies big and small are scrambling to be the word processors and spreadsheets of the big data era. The choice and different approaches can be daunting for those who are leading companies struggling with their own data needs. But few, if any, doubt that it’s time to do something to make data more valuable for workers throughout an enterprise.

The pace of competition won’t allow for the old ways, in which a mid-level manager comes up with a question he or she needs answered. Then it’s: Send the question to the analytics team and wait for an answer. By the time an answer comes back, the world has changed.

There is no question that we’re in an inflection point,” says Stuart Frankel, CEO of Narrative Science, a company that uses artificial intelligence to generate narrative summaries of data. “Where we started with the data, it was about data capture. That’s really what the data has been about — monitoring, tagging etc. There was almost this blind lust for data.”

The thinking, says Frankel, who appeared on the panel with Kahlow, was that if we just gathered as much data as we could, we’d figure out what to do with it, eventually.

“It turns out the collecting of the data and the aggregating of data is ultimately one step and the first step in what I think will be many” on the way to where the focus on data pays off, “because we can make better decisions.”

To look at the PC evolution analogy, it may be that we’re still in the early spreadsheet era when it comes to big data.

“We are just hitting the cusp of what is about to hit us and about what data can do,” Kahlow said. “Data is going to change everything that we do. And  every industry and every vertical is going to be disrupted. Today is a different world. I’m really excited to see what’s to come.”

And rest assured that it is coming.

Photo of Richard Frankel and panel by Mike Cassidy. Photo of IBM PC by Paul Sullivan published under Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.


Google Feud is addictive and educational — really

I stumbled onto Google Feud and now my life is over.

In fact, I’m writing about Google Feud because the only thing I’ve done since a co-worker pointed out Google Feud is play Google Feud. Write what you know, they say. Google Feud is now all I know. I even got a little panicky when it morphed into “bing feud” as an April Fool’s joke.

Screen Shot 2015-04-01 at 1.37.31 PM


You probably know about the Web-based game. In fact, you’re probably playing it now. Who isn’t? Patterned after TV game show “Family Feud,” the Web obsession offers players the first few words of a search query — for instance, “Tom Hanks is __” — and ask them to predict how Google’s autocomplete feature would finish the query. You’re allowed three misses as you try to come up with the top 10 answers, racking up points with each correct answer. (“Forrest Gump,” “married to,” “dead,” among others in the Tom Hanks example.)

A wrong answer delivers a bit red X, ala Family Feud. And yep, it’s pretty much addictive.

Silly, you say? Enlightening, I say. Think about those who create content for the Web and those who sell products on it. It’s challenging work. First, you need great content. But even then, you need to make sure that people can easily discover your great content.

There is plenty of work you can do filling content gaps, tagging products, creating paid search strategies etc., but meeting the challenge inevitably moves beyond human scale, in part because of the reality that Google Feud helps illustrate: Web users think of things and search for them in different and individual ways.

And so, Web operators turn to automation. The more automation, the better.

Consider, for instance, the wonders of auto-complete. It’s a key way to make a Web search easier for users. Type a few words, or even a few letters, and voilà, the phrase you were going for appears — thanks to the wisdom of the crowd that searched before you.

Can you imagine if a search engine tried to provide that feature manually? You don’t have to. In fact, Google in a spoof video imagined it for you with an interview of Michael Taylor, Google autocompleter.

But what if, instead of relying on machine learning, a person had to take an educated guess and list the full phrases users most likely intended to type? OK, what if I had to guess?

Who would build a time suck like Google Feud and why, you ask? That would be Justin Hook — and for fun.

“Google Feud’s surge of popularity really took me by surprise,” Hook said by e-mail when I asked him about his 15 or 20 minutes of fame. “It had actually been online for over a year, getting just a few hits a day, when one morning I woke up to find it on the front page of BuzzFeed. Thankfully, most of the feedback has been extremely positive. I’m not a programmer or game designer by trade. I just made the thing for a laugh, so I’m glad to find people are laughing along with me.”

And yes, I’m laughing. When I’m not pulling my hair out trying to come up with answers. Let’s just say, based on my track record, I’m no Michael Taylor.

I started with the “Names” category and was confronted with the partial query: “Sharon ___” I guessed “Stone,” which was a winner — top choice. Then Osborne, which yielded a big red X, ala the TV game show. I did slightly better when I tried “Osbourne,” which apparently is how Sharon spells her name. It was No. 4 on the list. And then I drew a blank, so I tried “Sharon Cassidy,” my cousin. Nothing.

I moved on to “Culture” and the partial query, “What the hell is a __” I went with my first thought: “Dongle.” Nope. “Cronut?” Uh-uh. How about “Wheatshocker,” given Wichita State’s recent March Madness run? Nope. Try: “hufflepuff,” “thot” and “hashtag.”

Really, people?

And I won’t get into the more complicated queries, like “My friend is addicted to ___” (I had no idea…) and “Should I stop___”  Why do I think those queries are coming from essentially the same set of people?

Anyway, that’s not important now. What’s important now is I get back to Google Feud and the next question that I’ll have no chance of properly completing.

Mike Cassidy is BloomReach’s storyteller. Reach him at; follow him on Twitter at @mikecassidy

Is data the new hiring manager?

If you simply skimmed the blogosphere and raced through a few news stories about trends in hiring and personnel, you might conclude that big data is taking the “human” out of “human resources.”

Not surprisingly, big data and data analytics are increasingly being deployed to figure out who to hire and how to best keep good employees happy while encouraging poor employees to hit the bricks.

NPR recently aired a story about companies like Pymetrics, RoundPegg and Knack, that are using brain games to measure aptitude, cultural fit and personality. The Wall Street Journal soon followed with a piece that talked about companies like Culture Amp and Ultimate Software Group that are working with big and well-known companies to identify valuable workers that some HR departments refer to as “flight risks.”


There is a lot at stake as companies work to hire the right fit and to hang on to their best workers. No one wants to spend money on recruiting, hiring and training only to have recent hires bailout quickly. And the cost in lost productivity alone is reason enough to keep track of which workers might be looking for other opportunities elsewhere. But those who understand human-relations data the best, say the current trend is not a wild departure from the sort of thinking hiring managers have always done.

“There is data that we use all day long, and historically, that we look at when we make a decision about whether to hire somebody — data around whether they have different skills, what their historical job history was; things that sit on the top of a resume,” says Jim Meyerle, who co-founded Evolv, a recently-acquired company that used big data to predict the best hires for hourly wage jobs. “That’s all data utilized in order to make a hiring decision that ultimately, hopefully, makes for a better match than just picking somebody at random.”

But as computers and software have become more powerful and adept at storing and sifting through data, more and more business practices and other activities are lending themselves to solutions that team up humans and machines.

In some ways, the examples of human resources professionals relying on human-plus-machine models point to the true value of data-driven systems: the ability to spread the power of data through a business or other organization. Some call it the democratization of data. Rather than requiring a hiring manager to ask a team of data analysts to work on a question and report back in days or weeks with an answer, new tools are moving toward virtually instant insights.

And as strides in artificial intelligence continue, the best tools will become smarter over time. They won’t simply report what is, but will offer predictive reports on what’s to come.

None of which cuts humans out of the equation.

Meyerle points out that when it comes to hiring, for instance, there is an endless amount of data that could be collected in order to make a decision.


“Part of that challenge is, what of that data is most relevant?” says Meyerle, co-founder of Evolv. Meyerle has moved on from Evolv, which was recently acquired by Cornerstone OnDemand. “There is a person that is an expert at determining what data we should be collecting and then it’s configuring or developing tools in order to capture that information in a systematic way.”

For instance, he says, it’s possible that in hiring for a given position, one particular data point is an incredibly strong predictor of the candidate’s success or failure. A good hiring manager might realize that on his or her own, but a machine is much more likely to use that intelligence consistently.

“When people are applying for positions, well, then you’re able to build that into the data model,” Meyerle says. “An individual hiring manager, an individual recruiter, might find something that is relevant and then forget about it or not keep it as a systematic thing that they’re looking for.”

A machine is also able to track how those factors lead to success — or not — over  hundreds, thousands or many thousands of hiring decisions and employees.

“The power of the machine is it’s enabling that person to collect what has been found across, ideally, a lot of different people who have been hired over time,” Meyerle says. “It’s helping enable that person to get the best data, the most relevant data, to make a decision on — because ultimately, the machine helps cut down on the data that somebody is looking at and delivers that in a digestible format to a person who ultimately needs to make a decision.”

The “human” in human resources is safe for the foreseeable future, Meyerle adds.

“I think at the end of the day, the human always has to make the decision.”

Photo of job fair by COD Newsroom and data center by Leonardo Rizzi published under Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.



Is “Undercover Boss” any match for big data?

When Michael Rubin was president of GSI Commerce he figured one way to get a ground-level look at how his sprawling enterprise worked would be to appear on the emmy-winning TV show “Undercover Boss.”

He no doubt gained some valuable insight masquerading as a working stiff for the popular CBS reality show in 2009. He also clobbered a GSI warehouse worker with a box the size of Manhattan while working in the trenches, a misstep that he later only half-joked had him worried about the injured worker suing the pants off the company.

Screen Shot 2015-02-27 at 10.47.32 AM

“GSI is done,” Rubin said, recalling his first thought before a crowd at the conference last fall. “We’re going bankrupt and the litigation is starting right now.”

He could have saved himself the drama.

We’ve reached a time when big data is the new “Undercover Boss.” Granted, terabytes of data being combed and crunched and crystallized in the cloud doesn’t make for riveting TV. Certainly not as riveting as the big boss messing up on the loading dock at GSI, which before it was purchased by eBay ran warehouse and shipping operations for e-commerce companies. But data can lead to some surprising and valuable insights of its own.

Rapidly evolving technology and data-handling practices have provided top executives with a view into their businesses unlike any that has been available before. Want to know who your customers are? Data can tell you. Want to know where to open your next store? Data has your answer. Want to know what isn’t selling and what is selling — or more importantly, what will sell? Data is your friend. Want to know who is working hard and who is hardly working? Again, data will tell you. Wonder whether employees are engaged and happy in their work? Data has got your back.

The stories of bosses learning by data, rather than by doing, are everywhere:

  • Yahoo CEO Marissa Mayer didn’t see the need to put on Groucho glasses and a wig to slip into the company cafeteria to find out whether employees who said they were working from home were actually working. Instead, she went to the data — the VPN logs that clearly showed that Yahoo’s work-at-home cohort was at home plenty, but not signing on to work nearly enough. The revelation led to Mayer’s 2013 edict that all Yahoos report for duty in person in the office. It caused, as you might recall, quite a stir.
  • Executives at The Fresh Market didn’t pose as grocery baggers in order to discover that they had been wrong about where their best customers came from. Rob Koch, vice president of real estate for the 170-store chain, told an audience at the National Retail Federation convention last month that it was data that blew-up execs’ notion of who they should be marketing to. When the company recently stepped up its data game, he said, it realized that its most valuable customer was not someone from the surrounding neighborhood who stopped in frequently. “Our average customer was actually there about once a month,” he said, “for a special occasion, which becomes a very different shopper to target, to market to, than somebody who lives nearby, who is there three times a week.” Rather than focus on getting three-times-a-week shoppers to shop five times a week, Koch said, The Fresh Market executives worked on getting monthly visitors to shop twice a month.
  • Chipotle executives turned to big data, rather than stealth, to keep an eye on its food supply chain. The fast-casual restaurant promises that its ingredients come from sustainable sources and animals that are naturally and humanely raised. The company has an intricate system of location and product numbers, bar codes and container codes that relies on cloud computing to keep track of just where the food it serves is coming from.
  • And there was no need for the C-suite dwellers at UPS to don brown uniforms and drive the boxy company trucks through suburban streets to figure out the best way for drivers to get from point A to point B and C, D and E. They looked to the staggering amount of data generated by orders, pricing, mapping and came up with the most efficient routes to run amid a big change in the nature of their delivery business brought on by e-commerce and the increase in deliveries to private homes.


It’s all part of a world in which bosses and line workers alike are increasingly embracing the idea that data wins arguments. Didier Elzinga, CEO of Culture Amp, a platform for employee surveys, says he’s seen the value that data and data analysis provide in the field of human resources.

HR professionals get into the field, Elzinga tells me, because they have a knack for understanding people and how people and organizations interact.

“The challenge that they often have, is they all know roughly what the issues are, but it’s very difficult to get everybody else in the company on board to actually do something about it,” says Elzinga, whose company is based in Melbourne, Australia. Data, he continues, “let’s that person actually mount an argument. It allows people to have a conversation at a much deeper level.”

In fact with data, the boss can dig pretty deep. Some companies are diving into employees’ inboxes and calendars to figure out who is wasting whose time. It’s a growing trend, called “people analytics,” according to The Wall Street Journal.

It’s clear that big data provides an incredibly powerful tool. We are fast approaching the point where CEOs will be able to deploy data in a way that leaves no stone unturned. And while the ability to access that sort of information will be a good thing for the business, there is no way it will ever be as much fun as watching the big boss assume a fake identity and flail away at a job he asks his or her employees to do every day.

Photo of Sucharita Mulpulru and Michael Rubin courtesy of the National Retail Federation and photos of Fresh Market by Natalie Maynor and featured photo of UPS trucks by US Department of Labor published under a Creative Commons license

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.



Another reason Twitter and UPS are poster companies for the digital age

If you had to pick two companies to symbolize the digital age, you could do worse than picking UPS and Twitter.

I can already hear the arguments: Twitter, sure. But UPS? Come on. The army of brown-clad drivers in brown-colored box trucks? What do they have to do with the digital revolution?


Pretty much everything. Ever since I read Tom Friedman’s “The World is Flat,” I’ve thought of UPS, once United Parcel Service, as a technology company. It turns out that the company that was all about parcels also knows a heck of a lot about packets — as in switching.

In fact, the company is a swirl of technology-enabled logistics — directing pizza trucks across country, moving most everything you can imagine from coast to coast and guiding those brown trucks down your street.

And now its latest technological initiative (free summary) as described in The Wall Street Journal: Orion, a computer platform developed to figure out the most efficient way to deliver packages in a shipping landscape that has become incredibly more complicated by the rise of e-commerce.

Orion sifts through the average of 120 daily stops that each UPS driver makes and comes up with the most efficient route out of a mind-boggling number of possibilities that takes into account customer preferences for delivery times, traffic rules etc. How mind boggling? The number of possibilities is a 199-digit number. For reals.

“Even if an optimal answer exists,” the Journal notes, “the human mind will never figure it out.”

But an algorithm will — and it should be a highly-paid algorithm. The charting chore, UPS says, will save the company as much as $400 million a year.

Twitter, of course, is the way we all communicate. OK, the way some of us communicate, some of the time. It has its own 2015 problem: So-called trolls who badger, harass and attempt to intimidate those who disagree with them.  And yes, Twitter’s solution, like UPS’s answer, involves algorithms.

But what’s notable about both these efforts, by both these companies that are synonymous with the digital age, is that neither is looking to algorithms alone.

Del Harvey, Twitter’s vice president of trust & safety, was fairly candid in a Q & A with The Wall Street Journal about the problem of harassment on Twitter. Maybe not as candid as CEO Dick Costello who wrote to employees that the company “sucked” at tackling abuse and that actually it had sucked at it for years.

Harvey acknowledged the difficulty in reining in abusive people on the Internet, but added that the company was working on the problem, including trying to figure out how to increase the role of law enforcement in the issue. Then The Journal noted that Costello said he intended to make it harder for Internet trolls to level abuse on users.  Then the news outlet asked Harvey about algorithms.

“WSJ: Can an algorithm solve this?

“Ms. Harvey: I don’t think that this is something that can be solved solely by software or solely by people. It has to be a combination of the two. We’ve made some improvements in terms of how we can process reports (of abuse). There is so much around context and intent. All those sorts of things make it really complicated (for an algorithm alone to assess) but (software is) absolutely something that we can use as an amplifier for the work that people are doing.”


And UPS? Sure, the company has invested 10 years and hundreds of millions of dollars in Orion, according to The Journal, but you know what? If drivers don’t want to use it, they don’t have to. They can rely on their own experience and intuition, especially when it comes to road repairs, accidents and judgment calls on the safest way to run the route. Though UPS apparently encourages drivers to use Orion.

It’s a tension reminiscent of the one between London cab drivers, who commit to memory the complicated streets of central London, and those advocating for GPS guidance.

Both Twitter and UPS are embracing a model that will no doubt be the accepted strategy for nearly every company in the 21st century: marrying the best abilities of the machine with the best ability of humans to come up with the best solution.

It’s an idea we wrote about recently in “Why Can’t We be Friends? The Case for Human+Machine.”  And it’s a theme that I can’t help seeing practically every time I turn around.

Photo of UPS truck by Atomic Taco and Twitter sand sculpture by Rosaura Ochoa published under a Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.



The rise of the machines comes with a big payoff

With all the algorithmically-driven convenience in our lives today, it’s sometimes hard to remember that we are still in the early days of a cultural shift that will see us increasingly interacting and building relationships with machines.

Sure, robots have been working side-by-side with humans in factories for decades. It’s not the machines that are new; it’s what the machines can do. Rapid advances in machine learning and processing power are producing machines that can perform jobs that the designers of those early factory robots had only envisioned.



And so naturally, we will be struggling with our comfort level as we get better acquainted with the machines in our lives. I won’t deny that stories like this Associated Press piece looking at the increasing number of jobs that robots will be doing can be anxiety-producing. From the AP story:

The Boston Consulting Group predicts that investment in industrial robots will grow 10 percent a year in the world’s 25-biggest export nations through 2025, up from 2 percent to 3 percent a year now. The investment will pay off in lower costs and increased efficiency.

Robots will cut labor costs by 33 percent in South Korea, 25 percent in Japan, 24 percent in Canada and 22 percent in the United States and Taiwan. Only 10 percent of jobs that can be automated have already been taken by robots. By 2025, the machines will have more than 23 percent, Boston Consulting forecasts.”

But I take a rosier view. Ideally, the trend of robots and machines taking on tasks for which they’re well-suited — repetitive jobs, jobs heavy on calculating, jobs requiring vast memory — will free up workers to tackle jobs more suited to human intuition and experience. The change won’t be seamless and it will require the rallying of education systems in countries like the United States to ensure that the human workforce is up for the upgrade, but overall it will be a win for the economy.

The key to success is marrying the best talents of humans with the best capabilities of the machines, as BloomReach wrote about recently in the eBook, “Why Can’t We Be Friends: The Case for Man + Machine.” Rather than leading to a world where machines take jobs from humans, I tend to see the future playing out more the way Cynthia Breazeal, of MIT’s Media Lab has described it.

Robots, she says, are all about personal amplification; about making humans better at the work they do; about being helpers.

Screen Shot 2015-02-11 at 4.37.12 PM

In the end, by taking factory jobs, robots could help build momentum for a manufacturing movement in the United States. True, the new manufacturing plants will have fewer humans on the factory floors. But manufacturing, and advanced manufacturing in particular, create other jobs beyond the factory floor — jobs in supply-chain management and operations, jobs in designing the systems that run the factory, jobs marketing and selling the products produced.

It’s a halo effect that I explored in a series of stories in the San Jose Mercury News a couple of years ago.

As I said, it’s early days and anxiety is to be expected. Change often brings anxiety. But it also brings progress — and in that regard we’re headed in the right direction.

Robot photo by Peyri Herrera  published under Creative Commons license, illustration courtesy of BloomReach.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

“Madden 15” Super Bowl win is bittersweet for one EA Seahawks fan

Wouldn’t you know that the most exciting Super Bowl in NFL history was the victim of a spoiler.

The New England Patriot’s last-minute win over the Seattle Seahawks provided a Hollywood ending, complete with spills (Jermaine Kearse’s circus catch), thrills (Malcolm Butler’s endzone interception) and chills (the requisite Gatorade bath for Patriot coaches).

The only problem: Electronic Arts’ “Madden 15” game called the outcome nearly a week before Super Bowl XLIX was played. The machine, it turns out, nailed it – right down to the final score.


“It just reminds us that the game is actually smarter than we are,” said Julie Foster, EA’s communication director. “It’s just a proof point that it’s the data that just drives the accuracy. We’re impressed by it ourselves, actually.”

The game’s predicting prowess is another example of the impressive results that can come of combining the best of humans with the best of machines. I’ve written a lot about that sometimes delicate balance lately, but I continue to run into example after example.

Now, a few things about the Madden example: It’s not like EA has turned to machine learning and human intelligence to eliminate a cancer or land a scientific device on Mars with unmatched precision. They were literally playing a game, after all. And no one is claiming that the Silicon Valley gaming company has come up with a reliable and repeatable way to flawlessly predict the outcome of the Super Bowl or any game.

But the EA folks clearly accomplished something that couldn’t have been done 25 years ago and it makes you wonder what we’ll be able to do 25 — or even five — years from now.

The feat wouldn’t have been possible, of course, without the help of humans. Sure, the humans on the football field. But also the programmer humans at EA who track what the football humans do and design algorithms to turn their real-world performances into a digital game.


“We have 300 plays per team, over 60 attributes per player that we update — and those are updated manually by humans here,” Foster said, “but it’s completely based on performance throughout the season.”

For the Super Bowl simulation, Foster said, the EA team goes to work immediately after the NFC and AFC championship games.

“We make sure that the rosters are updated accurately, that injuries are taken into account,” she said. “We run it with the very latest data that we have. Then we let the computer simulate it. No one is physically playing the game.”

And they play the game once.

So, how did the machine do?

Final score, 28-24. Check. Patriot’s quarterback Tom Brady throwing four touchdown passes. Check. Julian Edelman catching the winning fourth-quarter touchdown pass. Check.

The machine said Brady would throw for 335 yards. He threw for 328. It said Edelman would make eight catches for 106 yards. He had nine catches for 109 yards. It said Brady would be named MVP. Duh. He was.

And so, did anyone at Electronic Arts notice Sunday, as the commercials played on and Katy Perry performed and Super Bowl XLIX moved through the second half, that the real game was an awful lot like the pretend “Madden 15” game?

“I can tell you that tons of text messages started to fly in the fourth quarter between all of us, saying we might have a chance of nailing this thing,” Foster said of texts among her co-workers.

And yes, Foster was excited for her work team, but not so excited for her favorite football team. Yeah, the Seattle Seahawks.

“Now, being a Seahawks fan, the response I had was, ‘It’s not over fellas,’” she said of her own texts.

And then, of course, it was over.  New England’s undrafted rookie, Malcolm Butler, intercepted Seahawks quarterback Russell Wilson’s goal-line pass with seconds remaining in the game.

“As soon as the interception happened, it was like phones lit up all over, saying yeah, we nailed it,” Foster recalled. “If I had to have a silver lining as a Seahawks fan, that’s about as good as it gets.”

Really? As good as it gets?

“It’s not,” Foster corrected herself, “better than the win.”

The truth is, reality bites — even when it’s imitating art.

Madden 15″ screen shot courtesy of Electronic Arts. Photo of Zoltar by Jonathan Gleich published under Creative Commons license.

Mike Cassidy is BloomReach’s Storyteller. Contact him at; follow him on Twitter at @mikecassidy.



College football playoff teams are best selected by humans — not machines

When it comes to picking college football’s No. 1, humans reign

The long-running discussion pitting humans vs. machines is often steeped in data, science, social science, experimentation — and then the conversation rolls around to college football.

College football is about emotion, about becoming unmoored from reality, about ignoring the statistics and declaring, “We’re No. 1,” when, technically speaking, “We Are Much Closer to No. 137.”

College football is all about being like Phyllis the rabid Alabama fan and sports radio caller:

For the first time in years, the decision over who is the top NCAA football team in the country has been turned back over to humans. Yes, humans have always taken to the field of play to see who’s the better team, but since 1998 the declaration of who actually is No. 1, when all is said and done, was left up to a convoluted combination of human-generated polls and computer-calculated rankings.

This year it’s back to humans. This year, a committee of 13 has determined that the University of Alabama, the University of Oregon, Ohio State University and Florida State University are the four best college teams in the country and that they deserve to play a semi-final round, with the victors moving on to play for the national title.

The 2014 system marks a change from recent years when the top two teams playing for the championship were determined by a sportswriters’ poll, a coaches’ poll and a series of computerized rankings. It was a system that often produced howls of protest from fans of teams who thought their team ought have had a chance at the No. 1 ranking. Just as the system before that — relying on separate polls of writers and coaches to each select a champion — led to howls of protest (and sometimes to co-champions).

This new system? Yes. Howls of protest. And then some.


College football fandom is an emotional thing tangled up in school ties, bitter rivalries, nostalgia, regional jealousies and team loyalties passed down through generations. Given advancements in big data, processing power and machine learning, it seems entirely possible that a neutral group could devise a computer program that could come extremely close to flawlessly selecting  the best team in the country.

“But what fun is that?” Asks San Jose Mercury News sports columnist Mark Purdy. He even has an answer: None.

“One of the reasons that college football has been a favorite sport of mine, is the history” says Purdy, who’s covered a variety of sports over a distinguished career. “College football has been all about humans assessing what’s good and what’s bad. I think if you have a correct number of people assessing these games, I think you end up with a pretty good consensus of what’s the best team.”

College football presents some special challenges: Most schools belong to conferences, which dictates the bulk of their schedules. Head-to-head matches involving top teams are often not available, so producing rankings based on what happens on the field is often impossible.

That, everyone agrees on. It’s what to do about it that sparks debate. Austin Narber, who blogs about sports — Iowa State University basketball in particular — has written that machines are the way to go. From his post earlier this year:

“There is one fact you cannot deny: the human brain, no matter how intelligent its possessor appears to be, cannot begin to fathom the amount of information that goes into calculating things like tempo, adjusted field goal percentage, turnover percentage, luck, or strength of schedule. It cannot fathom how these statistics are ever-changing with every single possession of every single game, every single day of the year. By not supporting the existence of the computers that flawlessly calculate things like this, you are instead supporting the idea that human error is acceptable.

Let’s say you’re answering a trivia question worth one million dollars. You have two lifelines:

  1. You can ask a group of your relatively intelligent friends to collaborate and come up with an answer, or
  2. You can Google it.

Don’t sit there and tell me you wouldn’t go to the computer.”


But maybe part of the fun with picking a national football champion is the quirky uncertainty. Maybe having humans involved in picking the four teams that will have a chance to win the national championship just adds to the delight. Humans are flawed, unpredictable. They do screwy things. Sometimes it’s fun to watch.

“Football is unique,” Purdy says. “There are a lot of teams; all these different cultures of these teams. There are leagues from coast to coast, where teams don’t play each other. So you kind of have to rely on the eyeball.”

So why not pick enough eyeballs — say 26 in the case of the current committee, attached to people who know something about football, and let them set up the playoff?

“It’s football,” Purdy says. “It’s not deciding which general should defend us in a nuclear war.”

And that’s just it. I’ve written a lot lately about the balance between humans and machines. I’ve looked at Yahoo’s early efforts to categorize the Web. And London cabbies’ ability to navigate the streets of central London based on their knowledge and experience. I’ve written about centaur chess players who team up with machines to become the best possible chess players they can be. Mostly, I’ve written about working to get things done.

But what if, like in the case of football, human foibles were core to the entire exercise? To take it to the extreme: If you’re going to rely on machines to pick college football’s national champion (based, of course, on wins, losses and a long list of other metrics), why not rely on a machine to determine the outcome of each individual game and do away entirely with the need to play them?

Why not? Because it would be silly. And a whole lot less fun.

Or as Purdy puts it:

“The college thing I think is, more humans equals more fun. Always and forever.”

Photo of Illinois vs. Michigan football by Adam Glanzman and Rose Bowl at night by Navin75

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Centaur chess brings out the best in humans and machines

The story of IBM’s Deep Blue computer defeating world chess champion Garry Kasparov in 1997 has been told so many times that it’s practically shorthand for the philosophical debate over man vs. machine.

But the story lacks subtlety and perhaps the right moral. Deep Blue was only the beginning; and out of Kasparov’s defeat grew a type of chess player that more richly illustrates the interplay between man and machine in 2014: All hail the centaur.

Yes centaur — and Kasparov was apparently the first. Rather than half-horse, half-human, a centaur chess player is one who plays the game by marrying human intuition, creativity and empathy with a computer’s brute-force ability to remember and calculate a staggering number of chess moves, countermoves and outcomes.

The centaur story is an elegant example of the way visionaries see the optimal interplay between humans and machines. Teaming the two in chess, experts say, produces a force that plays better chess than either humans or computers can manage on their own.


Centaur chess is all about amplifying human performance.

The human plus machine style of play is called Freestyle (often played online) and the rules of the game allow chess players to consult outside sources — books, other humans and most importantly computerized chess engines that are stocked with the moves and results of thousands or millions of chess games that have been played through the years. The Freestyle games are timed, forcing players to think on their feet while managing the clock.

“You merge the computer games with the human games and you’ve got something that is quantitatively and qualitatively out there,” says Nelson Hernandez, a suburban Washington D.C. data analyst who is a member of one of the most successful Freestyle teams in history.

The three-member team is led by Londoner Anson Williams, a reserved engineer and software developer, and is rounded out by Williams’ girlfriend, Yingheng Chen, a math whiz who has become an expert at analyzing Freestyle chess.

None of the three consider themselves accomplished chess players and with the help of the machine they needn’t be. In fact, you could argue that chess experts — grandmasters — are at a disadvantage when it comes to Freestyle. It isn’t that the same computer-based fire power isn’t available to grandmasters. It’s that they sometimes fall prey to the very human belief that as experts in their fields, they know better than the machine.

George Mason University professor Tyler Cowen covered the syndrome in his book, “Average is Over.” In it, he tells of U.S. grandmaster Hikaru Nakamura, “who was not a huge hit when he tried Freestyle chess. His problem? Not enough trust in the machines. He once boasted,” Cowen continued, “‘I use my brain, because it’s better than Rybka on six out of seven days of the week.’ He was wrong.”

Rybka is a top chess engine, part of the machine.

The flip-side to Nakamura’s hubris, of course, is placing too much trust in the machine. Nicholas Carr, author of “The Glass Cage: Automation and Us,” has said humans risk being seduced by automated programs because they carry out work in a “black box.” Questions in; answers out.

But centaurs understand that the machine has limitations. Chess is too complicated a game, with too many possible moves leading to a slew of other possible moves, for today’s computers to solve. What they can do is push players toward perfection by charting out a series of flawless opening moves and then providing recommendations the rest of the way.

Not all machines are created equal, however. And different machines will provide different answers to the question: What’s my best next move? Then the human — who by the rules of Freestyle must physically make the move — has to decide not just what’s the right move, but what’s the right move against a specific opponent at this particular point in the match.

“That’s where someone like Anson really shines. He’s augmenting the computer with his own native abilities,” Hernandez says. “A chess player has got a certain set of cognitive equipment. An almost superhuman memory is what’s required, a tremendous ability with spacial concepts and to be able to look ahead, a great strategic sense of when you should exchange this piece for that.”

Think of it as something like a trip to the moon: A pilot has the skill to land a spaceship on the lunar surface and he or she can use his or her eyesight and human experience to determine whether a landing area is suitable. But the pilot would never be in the position to be making those decisions without a big boost from the rocket that got him or her to the moon’s orbit in the first place.

2346661035_8ffd3c5703_z (1)

The rocket for Team Anson Williams (The trio plays in different tournaments by different names to make it harder for opponents to scout them.) consists of the team’s massive database of tens of millions of chess games, called an opening book, that have been played by computers and humans — every move, every outcome, billions of unique chess positions.

Hernandez says he’s spent over 10 years compiling the data, which he says the team uses to rank the best moves after analyzing the game situation with custom, proprietary software.

“I hardly ever take a day off,” he says. “I have this incredible talent for dull, repetitive tasks. Who else would spend 10 years of their life, every single day, collecting chess games for their database? It’s almost borderline crazy.”

As much as he loves data, Hernandez is well aware that as with data in so many fields, it’s not just having lots of data that matters, it’s what you do with it.

“What Anson does, probably as well or better than anyone else, it’s his secret really, is you need to be able to process real-time data sets more efficiently, more rapidly than anyone else. That’s a cognitive skill. You need incredible nerve, really. You’re under a clock playing against someone else with a similarly powerful computer array and the slightest error is almost invariably punished.”

The combination of Williams’ cool under pressure and the machine’s vast catalogue of chess moves works. Cowen’s book recounts a four-tournament run in which Williams’ team won 23 games and lost one while playing 27 to a draw.

Williams did not respond to requests for an interview. Reserved, remember?

So which is more important in the world of centaur chess? The human or the machine?

“It’s absolutely more important to have good hardware and software,” Hernandez says. “In Freestyle, we have competed against grandmasters two times and we have defeated two grandmasters. And we’re not ranked chess players, you know. To me, that tells the story. Is chess knowledge useful? Of course it is. It’s not that it’s useless. I’m just saying that if that’s your main strength and not a lot of other things, you’re going to lose. It’s simply not enough.”

In other words: Back in the Deep Blue days, at least Garry Kasparov on his own had a fighting chance. Today, not so much.

Photo of chess pieces by Ina Centaur, chess board by Kamil Porembinski and centaur by Anders Sandberg published under Creative Commons license. 

Mike Cassidy is BloomReach’s storyteller. You can reach him at; or follow him on twitter at @mikecassidy.



How humans and machines can make nice and get stuff done

Maybe the secret to the complicated relationship between humans and machines comes down to a simple instruction: We all just need to get along.

The conversation has been a hot one since the dawn of the Industrial Revolution: Would machines take our jobs, rob us of our creativity and intuition and in the process turn us into drones that are more machine-like than human?

But when everyone simmers down, it becomes clear that this is not an either-or situation, that in fact taking the best of both — the best of humans and the best of the machines — can lead to accomplishments that neither could achieve on its own. The truth is that in the end, machines amplify human potential.

“When you are building this whole man and machine model, the machine needs to be a companion,” says Ashutosh Garg, BloomReach’s chief technical officer. “Machines need to be an assistant. You don’t want to be competing with a machine.”


Garg, a former IBM researcher and Google scientist, says that machines and humans are good at different things — meaning the potential for partnership abounds.

A machine, for instance, is good at remembering where your keys are; you are not. A machine can churn through massive amounts of data and come up with a result. You would be overwhelmed by massive amounts of data. But you are good at generalizing based on past experience. Machines aren’t. You’ve got intuition; machines don’t.

The differences between humans and machines mean that each has a role. For instance, Garg says, think about innovative ideas and the evolution of them.

“As a machine, you can come up with an algorithm that will make your iPhone better,” he explains. “But a machine cannot come up with an iPhone. The machine is going from the iPhone 4 to the iPhone 5.”

In other words, innovation takes a creative spark; a very human creative spark. But a machine is able to amplify that creativity. A machine can in essence survey millions of iPhone users and track the performance of millions of iPhones. It can tell human engineers, designers and marketers what apps iPhone users favor and what apps they are ignoring. It can tell the humans at what times and for what purposes people are using their iPhones.

Are consumers primarily using their phones to listen to music? The humans are going to want to work on improving the next version’s speakers and acoustics. Have iPhones become consumers’ go-to cameras? Maybe it’s time for designers and engineers to improve upon the lens and flash. Is the iPhone battery dying at dinnertime instead of bedtime? Time for a better battery. Are users complaining about the weight of the phone? Machines can process millions of comments and figure that out. Then the humans can figure out how to make the phone lighter.


Or think of the human and machine partnership in launching a marketing campaign, Garg says. An experienced marketer, for example, might rely on intuition and experience to conclude that blue sweaters are going to be a big seller in the coming season.

“Machines cannot come up with that. The amount of data they would require to come up with that is practically non-existent,” he says. “Anytime you are trying to do an innovative thing, out of the box, trying to set a trend, you don’t have the data for that. You have your wild intuition.”

But if you’re building a results-oriented marketing campaign to sell those blue sweaters, turning to a data-infused, intelligent machine makes a lot of sense.

“If you’re doing certain things, machines can help you improve things — a lot,” Garg says. “Machines can tell you the likelihood of this campaign performing, of having conversions, what the cost is.”

A human marketer can figure out what kind of Web page makes sense for the campaign — a theme page, a splash page, a collection concept page. A machine can quickly process a ton of product data — what are consumers searching for, what do they buy as a result, what other products do they view and purchase as a result of their initial search — and determine exactly which products should go on the page. The machine can not only come up with the answer faster, its answer is more reliable than one arrived at by a human, who has no chance of processing the same amount of data and who is left to rely on partial data and artistic hunches.

“Optimization is a tedious task. Machines are very good at that, churning through a bunch of data,” Garg says. “Machines are great optimizers and humans are great innovators.”

It would seem the surest path to peace and prosperity when it comes to the co-existence of humans and machines, then, is for everyone to keep that squarely in mind.

Photo of Ashutosh Garg by BloomReach; photo of Lego robots by Nick Royer and factory worker by Kheel Center published under Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Microsoft’s Cortana is an NFL game-picking machine, but still a machine

Deep Blue sounds like something it would be OK to get beat by. Even Watson, IBM’s answer-in-the-form-of-a-question brute, has a name that sounds like a winner.

But Cortana? Come on. It sounds like a Chrysler sedan out of the 1980s. But it’s an intelligent machine; Microsoft’s personal assistant in fact, the tech giant’s answer to Apple’s Siri.

And it beat me. And my wife, Alice, too.

Maybe you’ve read about Cortana’s sports prognosticating prowess. With the help of Microsoft programmers and the company’s Prediction Lab, the Windows phone digital assistant nailed its World Cup soccer predictions, calling 15 of 16 matches correctly. So naturally, Cortana turned its game-picking attention to the kind of football you don’t play with your feet — the National Football League.

The Microsoft machine has been predicting the outcome of every NFL game since the beginning of the season. It’s 69-37 or calling it right 65 percent of the time. I couldn’t resist. I had to go head to head with the machine.

I’ve been exploring the intersection of humans and machines and I couldn’t just stand on the sidelines. (Get it?) I decided, starting with Week 7, to go head-to-head with the disembodied voice. And for good measure, I dragged my wife, Alice, into the competition, given that she actually knows something about NFL football.


Through six weeks, Cortana had proven itself a worthy opponent. Me? I keep an eye on football through the season, but I tend to focus on a couple of teams — my beloved Chicago Bears and my adopted San Francisco/Santa Clara 49ers. On the other hand, Alice, a lifelong 49ers fan, will watch anybody, and everybody, play — and not just watch, but fully dedicate herself to the unfolding drama. She knows football.

Cortana crunches data — wins; losses; artificial turf or grass; home or away; weather; strength of schedule; team statistics. Then it adds in public sentiment gathered from Facebook and Twitter, an attempt to inject some human wisdom into the process. The idea: If the “crowd” is really big on a team or really down on a team, it could be an indication that something is going on that isn’t reflected in the numbers.

It’s telling, isn’t it, that Cortana’s programmers would find a way, albeit a rather inexact way, to add the human element. I’ve written before about how the man vs. machine construct is outdated. Instead, it’s best to think of the ideal as human + machine.

In fact, it’s also significant that Microsoft turned to a team of humans to make their machine feel more human by mastering the art of chitchat. Now that Cortana is picking football games, they might want to move from chitchat to trash talk (maybe in the next release).

But back to picking NFL games. There are a few caveats. For one, Cortana doesn’t factor in the point spread when picking winners. Maybe that adds a level of complexity that the system isn’t ready for. (Microsoft declined an opportunity to talk to me about the thinking behind Cortana picking NFL games.) Gamblers typically can choose either an underdog, which is spotted points, or a favorite, which gives up points. If the point spread is 6.5, for instance, the favorite must win by seven or more for a bet on the favorite to pay off.

The failure to factor in the point spread left one professional bookmaker completely unimpressed with Cortana’s 65-percent performance.

“I could do that,” he said. In fact, he added, a chicken could do that. Just put the chicken on a paper with the games listed and watch where it pecks. It’s all about the spread; and picking a game with a spread attached takes some significant processing power or brain power.

And it requires something more; something that humans bring to the table and machines just can’t, says the bookie, who didn’t want his name published. (Can you believe it?)


“You have your empirical data, which is out there. There is so much data, so many statistics on sports. You could rate so many columns of data, coming down to anything,” he says, including which team wins when a given announcer is calling the game. “The thing that you can’t do, is essentially know what’s happening, to have that inside information. It’s that last little tidbit of information that skews (things). That’s the true element that tips the scale.”

Little things, like knowing which player’s marriage is on the rocks; or who’s battling addiction, are sometimes the keys to the game that don’t show up on television graphics, the bookie says. And the only way to learn about that stuff is to build relationships.

Yes, build relationships, the way humans do.

On the other hand, a machine isn’t going to pick the Bears just because they’re the Bears, which is what I did in my Week 7 contest against the machine. (Dolphins 27; Bears 14.) And Cortana isn’t going to take Cleveland over Jacksonville, based on a long-held bias against expansion teams. No, that would be me. (Did I mention that Jacksonville won 24-6?)

OK, so how did the humans do? I know you’re dying to find out. If you must know, Cortana and I were tied at 10-4 going into the Monday night Texans vs. Steelers game. Alice was at 8-6. Hey, she’s human.

And in the end, so am I. Cortana picked Pittsburgh in the Monday night contest. And wouldn’t you know it? The Steelers won by virtue of a late, second quarter scoring blitz, racking up 21 points in 73 seconds.

Who knew? Cortana apparently. But hey, I gave the machine a run for its money.

The only catch: As with any machine learning system, Cortana is no doubt getting smarter as the season goes along. Me? Not so much.

Cortana image courtesy of Microsoft

Mike Cassidy is BloomReach’s storyteller. Contact him at; follow him on Twitter at @mikecassidy.

Yahoo’s handmade directory no match for the machine

A Silicon Valley landmark is coming down at the end of the year and while landmarks in the tech-centric region are sometimes more cyber than brick-and-mortar, their passing is still a cause for reflection.

In this case, the landmark is the Yahoo Directory, an old-school, hand-crafted tool to search the Web. Admittedly, in this time of algorithmically-powered search engines, the 20-year-old directory feels a little like turning to a telephone switchboard to complete a call you could make on your iPhone 6.


Hand-crafted — that’s how they did it back in the day. Yahoo employed an army of “surfers” who combed through digital piles of URLs, submitted by site owners and the public, and decided which were worthy of inclusion in the directory and under which categories and subcategories they should be listed.

“It was pretty wild,” says Steve Berlin, Yahoo employee No. 14 and the company’s first full-time surfer. “Basically, everyone was given a list of hundreds of sites and every day they were given a new list or every week they were given a new list. Everyone had their own specialties.”

A music fan might be in charge of vetting and categorizing new music sites that were submitted by their developers. A book-worm would categorize books. A sports nut might sort out sports teams and fan sites.

“Since we were from all over, you know geographically, we’d all add sites from our own geographic area that we knew better than some random person off the street,” says Berlin, now a Massachusetts resident who surfed at Yahoo until 2001.

It was as if the surfers were building the knowable Web by hand. They had rules: A website had to be substantive, no thin content. A site needed to be a site, not just a page. And they had standards, which was where Erik Gunther came in. Gunther worked at Yahoo from 1998 to 2008 and his job was to deal with website owners who wondered why their site was not listed in the directory.

“I dealt with people who were eager to get their site listed back when it was a big deal,” says Gunther, a San Jose resident who writes news for the website. “It was a getting listed could make or break your business kind of thing.”

Organizing the Web through a hand-made directory is hard to imagine today. But keep in mind that the system, conjured up at Stanford University by company founders Jerry Yang and David Filo, was better than what was out there at the time.

And it felt like the start of something big — the start of teaming bright minds with powerful machines to get the work done. With the help of human editors, the directory created neatly organized pathways that would guide users to the information they wanted or needed. Looking for the weather forecast? Click on “News & Media,” followed by “Weather,” then “By Region,” followed by “U.S. States,” then “California,” then “Cities,” “San Jose,” “Weather Bug” and “Mountain View.” Done.

Somehow the description “simpler time” doesn’t seem to apply. But it was a time that many look back on with fondness — a sort of digital dawn. I recall visiting Yahoo in 1996, back in the directory days. Founders Filo and Yang greeted me shoeless and showed me to their desks, complete with requisite half-eaten burritos and futon mattresses for under-the-desk nap sessions. A Yahoo employee rode by on a bike (indoors) as we talked and a whiffle ball game broke out in a hallway.

Danny Sullivan, a founding editor of Search Engine Land, wrote a post last month chiding Yahoo for its quiet send-off of what the piece’s headline said was once the Internet’s most important search engine. (Hat tip to Sullivan for publicizing the story, which was mentioned in one paragraph of a longer post on Yahoo’s corporate blog.)

“I think Yahoo has grown and changed so much that it doesn’t even remember or respect its own history, perhaps because there are few left who recall it,” Sullivan said in an e-mail response to my question about Yahoo’s very low-key sendoff. “That’s a shame. It’s also because Yahoo simply might feel calling attention to the closure is somehow a failure; so it doesn’t want to be seen as celebrating a failure. But the directory was so important, so foundational to Yahoo, that this should have overridden those other concerns.”

For her part, Mashable’s Christina Warren, remembers the thrill of having her first website added to the directory in 1996 and shares Sullivan’s criticism for the shrug with which Yahoo is shutting it down.

Yahoo has never been one to mourn the passing of its own history, a corporate characteristic I wrote about when it shut down its iconic billboard off the approach to the Bay Bridge on the edge of San Francisco.

And the truth about the directory is that like those operators working the switchboards of the last century, search technology has by-passed the Yahoo Directory. The Internet was a very different place 18 years ago. As early surfer Berlin told me:

“If they wanted to keep it up, no one would have enough money to throw the manpower at it, that would be needed to do that.”

Think about it: While the exact numbers are open to debate, Internet Live Stats reports that there were about 258,000 websites when I visited Yahoo in 1996. Last month, the number of sites passed one billion. So if, as former surfer Jon Brooks reported on KQED radio, Yahoo needed 100 workers at the time to keep track of the World Wide Web, the Sunnyvale company today would need about 388,000 employees to do the work.

And what would it do tomorrow?

Sure, Berlin and Gunther are both sad that the directory will be no more — sad maybe the way you would be to see the old house where you grew up torn down.

“It was the best job I ever had and the best job I ever will have,” says Berlin, who now does computer consulting. “They were incredibly great people to work with. They were all very smart and clever people and we knew that we were on to something special here. Basically, without Yahoo the Web itself would not be around the way we know it right now.”

But no, Berlin says and Gunther agrees, there is no logical argument for keeping such a manual process going in a world where the Web is vast and growing and algorithms are able to pull most of the weight in keeping track of it.

Clearly, in the case of the Yahoo directory, the work balance between humans and machine had become seriously out-of-whack. It is the natural course when technology evolves. But it in no way diminishes the value of human beings or signals their coming obsolescence.

Instead, it’s a lesson that some things are better left to the machine, while human minds are freed up to work on creative solutions to the next big problem.

Photo of switchboard operators by Seattle Municipal Archives published under Creative Commons license, Yahoo 1996 screenshot courtesy of the Internet archives.

Mike Cassidy is BloomReach’s storyteller. Contact him at mike.cassidy@bloomreachcom; follow him on Twitter at @mikecassidy.

A machine is no match for The Knowledge of London’s cabbies

Who wouldn’t be rooting for London’s black-cab drivers in the latest assault on tradition and perhaps the drivers’ livelihoods?

These drivers, who find themselves fending off ridesharing service Uber, spend years learning their craft before they actually earn the rank of London cab driver. They know not only how to get from point A to point B in central London, but also how to get from point A to points C,D,E,F,G and from point G to Point K or point Q to point L. You name it. London’s black-cab drivers know the six square miles radiating out from Charing Cross better than they know themselves. They know every street and alley, every monument, historic site and landmark — roughly 25,000 streets and 20,000 monuments.

The drivers’ geographic brilliance is known as The Knowledge and every black-cab driver must demonstrate that he or she has mastered it before being awarded a taxi medallion (or a green badge in this case). It’s a mind-boggling amount of information that takes on average four years of walking and driving the streets of London to learn.

And yet, it’s something your smartphone knows the day you take it out of the box, right?

Not so fast, says London cabbie David Styles, who writes the CabbieBlog.

“Satnavs are hopeless in London’s complex and ever-changing road network,” Styles tells me by email, referring to satellite navigation systems. “Often an alternative route might be longer, but faster. In fact, many of our customers will have a preferred route.”

And I want to believe Styles — and admit it, you do, too. I want to believe that there is a benefit to being driven by a man who spent more than four years studying the streets of London, rather than a driver who takes machine-issued directions well. There is something intuitive about the argument that there are subtleties, aberrations, emotional factors, that humans can understand and machines can’t — the idea, for instance, that the longer route is actually the shorter route.

Cabbie Rob

Cabbie Robert Lordan, who blogs at View from the Mirror: A Cabbie’s London, tells me the times when human intuition is invaluable in his black cab are endless. Take Jermyn Street, pronounced “German Street.” How many enter that correctly into a GPS device? He’s been flagged down by people needing to get to the emergency room, who figure a black cab is faster than an ambulance.

“I’ve lost count of the times in which a passenger has mixed up roads and destinations,” Lordan adds by email. Like the woman who wanted to go to King William Street in Covent Garden. He knew there was a King William Street, but it’s in the financial district. And he knew that there was a William IV Street in the vicinity of Covent Garden, so he asked the woman where in particular she was headed. A restaurant, she said. Terroirs, he asked? Bingo.

“I frequently meet passengers who are unsure of where they’re going, who are only vaguely aware of the address, who get muddled up,” Lordan added. “People who want to go to multiple destinations (and not always by the most direct route); people who change their mind halfway through a journey; people who want hotel, restaurant, pub recommendations; ideas of places where they can take their kids; people who want to know about London and its history; about obscure museums and so on and so on.”

But we’re steeped in a world of technological wonders. We’ve seen what machines can do, how they can make life easier, tasks quicker and businesses more profitable. It sets up a classic debate: human vs. machine.

Ironically enough, in this digital age, that binary view doesn’t do the discussion justice. As I’ve written before, often it’s the case that the right answer is human plus machine. Big data is a fantastic tool for rooting out problems, predicting the future and designing the best course of action. But it takes humans to interpret that data and to design the right questions to ask, and the right methods to go about answering them. It takes an approach that yields what has sometimes been described as thick data.

And it turns out, thank goodness, that there are demonstrable things that a London cabbie armed with The Knowledge can do that the GPS guiding an Uber car through London cannot.

Start with the fact, uncovered by neurologists at the University College London, that the hippocampi of London cabbies are bigger than those of the general population and that their memories are better, too.

“The city becomes a part of you,” Lordan says, offering a plausible non-scientific explanation. “You have a handle on it and know how to deal with the many problems and conundrums, which occur every day.”

And then consider the thoughts of Paul Densham of University College London, who is not a neurologist, but a geography instructor whose work has been incorporated into modern surveying and positioning systems. There are times, he says, when an algorithm is not up to the task. What does an algorithm know about timed and untimed traffic signals, or the events that cause pedestrian traffic to surge and repeatedly halt auto traffic as mob after mob avails itself of the crosswalk?

“There have been attempts to bring in real-time traffic feedback,” says Densham, who is a reader at UCL, “but again, if everybody is getting the same feedback, then all our algorithms are going to recommend similar diversions. The irony is that at the moment, the black-cab driver is probably in a better position than the Uber software in terms of working in London.”

In other words, the cab driver can find the longer route that is shorter. It’s what they do and in doing so for as long as any Londoner can remember, the black-cab drivers and their distinctive cars have become an enduring thread in the city’s fabric.

“They’re like the red bus and the red post box and what have you,” Densham says. “They’re sort of an icon of Britain and London in particular.”


Which is not to say there aren’t circumstances in in which a black-cab driver might benefit from teaming up with a machine.

Styles pointed out that cabbies use a system called Hailo and GetTaxi that allow passengers to order up a black cab, rather than hail one on the street. A new service called Maaxi has recently launched. And a driver who asked that his name not be published, said he turns to GPS to navigate outside of London — never in the city center, where he says he knows every restaurant, ATM, nightclub and toilet.

Toilets? People ask to be driven to toilets?

“Yeah. One guy needed one and I got him there,” the driver says.

Densham says he could see a day where a machine becomes the black cabbies’ ally even in the center of the city. Say, for instance, he says, the already horrendous congestion in London becomes worse. It might become necessary for cabbies to have more technology to help them overcome the jams.

Though don’t count on Styles firing up a computer to help him find his way around a bottleneck. No need, he says, to let a GPS system help with the route, so he can chat with passengers or describe points of interest.

“You know, it takes us a few seconds to work out the route in our heads before we drive off,” he says. “The rest of the time, we tend to be on auto-pilot. An old cabbie once said to me that you know the time when you have cracked it is when you don’t have to think anymore.”

Styles offers a fairly convincing argument and some reassurance that the black cabbies are not going away anytime soon — a notion I’m happy to embrace.

In fact, let’s just say it: All hail the black cabs of London.

Cover photo of London cabs by Sammy Albon and cab photo by James Barrett published under Creative Commons license. Photo of Robert Lordan courtesy of Robert Lordan.

Mike Cassidy is BloomReach’s storyteller. Contact him at mike.cassidy@bloomreach; follow him on Twitter at @mikecassidy. BloomReachers Justin Fogarty and Stella Treas in London contributed to this report.

Whether it’s baseball or retail, combining experience gained and data collected is the path to success

Let’s face it, Major League Baseball and those who follow it like tradition.

Baseball is all about poetry in motion, the cut of the grass, the timelessness of the game, the historic heroes and the manager’s gut. Sometimes, a manager just has to go with his gut. He knows a pitcher in the bullpen owns the pinch-hitter coming to bat. He knows the wheel play is on and a bunt isn’t the wisest course just now.


But baseball, like all professional sports, is about something else, too. Winning. And it turns out one way to increase the chances of doing that is to turn away from the gut and toward the data. Oakland Athletics general manager Billy Beane and author Michael Lewis did more to promote that thinking than anyone, as outlined in “Moneyball: The Art of Winning an Unfair Game.” (The view, of course, applies to other businesses, as well.)

I was reminded of this over the weekend while listening to baseball guru and KNBR-AM radio host Marty Lurie and his interview of Arizona Diamondbacks’ coach Mark Weidemaier, who was talking about a baseball hot-button: The practice of employing an infield shift to better defend against left-handed pull hitters. The move, in which three defenders (or four in extreme cases), take positions between first and second base, appears to be working.

And it makes sense: Left-handed hitters naturally hit the ball to the right side of the infield. But Weidemaier said he’s not going with his gut when he sets the Diamondbacks up in a shift.

“I’m in charge of putting together the defensive alignments, our optimal positions, where we’re going to play to start a ballgame against each hitter,” Weidemaier, who’s been in the game for three decades, told Lurie. “We look at all the spray charts. We look at all the data.”

He added that looking at charts that show where on the field hitters hit the ball is nothing new. Coaches and managers have been jotting that down for decades. But something has changed since the days of dugout clipboards.

“Now you have the computer, which gives you these charts at a click of a finger for 4,000 at bats,” he said. “When you see the shading, when you see the charts, it’s wise to take advantage of the data.”

It is wise, but not necessarily universally accepted by fans and pundits. The idea that stats can outsmart baseball brains is the source of a heated and ongoing debate. How can a machine outperform years of playing and coaching ball?

(Turns out that some wonder whether the shift should be outlawed, while others believe the game will right itself as it always has.)

Sound familiar? Whether the field is education, climatology, politics, investing, journalism, retail or baseball, there are people involved, people who develop expertise and experience. People, who at times, trust their guts.

Chicago White Sox announcer Ken “Hawk” Harrelson, for instance, isn’t real big on data as he explains in this video debate with the MLB Network’s Brian Kenny.

And maybe these sorts of all-or-nothing debates unintentionally make the most important point: The answer is not to give over all decision-making to the machine. Nor is it to completely disregard the data and go with the gut every time. The secret, as John Kay points out in the Financial Times, is to combine the best of both types of intelligence to give an organization the maximum chance to succeed.

Lurie, who’s as good as they get when it comes to baseball knowledge, touched on the point with Weidemaier, when he pointed out that in order for the shift to work, the team’s pitcher needs to put the ball in the right place to ensure the batter indeed hits it to the right side.

“To do it in harmony,” with the pitcher, “is the best way to go about it, obviously,” Weidemaier said. “We try to put our guys in the optimal position to begin with and then work off that as for how our pitchers attack hitters during the game. And I keep charts during the game myself. As the year goes on, I build off of that.”

And as he builds profiles of hitters, he shares them with the game’s pitcher and catcher (they meet before every game) and the team’s defensive players (they meet before every series) to ensure that the human element is very much a part of the equation.

It sounds like Weidemaier has found the sweet spot: Let the machine do the grunt work of tracking thousands of at bats, while he puts his considerable experience to work helping pitchers and defenders understand what it all means for them.

At that point, the debate is no longer about which is better, the human or the machine. At that point the question is how can one best improve the insights provided by the other.

Ball park photo by viviandnguyen and mitt by Andrei Niemimäki published under Creative Commons license.

Mike Cassidy is BloomReach’s storyteller. Contact him at and follow him on Twitter at @mikecassidy.