old car near field

Have you considered all the costs of your legacy site search system?

old car near field

Parts one and two of our series on the path to better site search looked at the difference between hosting systems onsite versus in the cloud and the difference between having semantic understanding capabilities and not having semantic capabilities.

But what, exactly, is the bottom line when it comes to these differences. The bottom line, it turns out, is the bottom line. In the end, you see, the difference between onsite and cloud, and semantic and not semantic, could be the difference between profit and loss.

A legacy site-search system can be a bit like a cool, old car. It’s powerful, able to get the job done, but it takes a lot of care and feeding to keep it running well. Despite the car’s charms, the cost of ownership is not a pretty sight.

Rules-based site search engines like Oracle’s Endeca require a lot of manual labor. Unlike search engines powered by machine learning and artificial intelligence, rules-based search engines need to be told what to do in particular situations.

Take synonyms. If you want your Endeca search engine to know that “running shoes” can be the same product as “sneakers,” you need to tell it. Or consider gender. If you want your rules-based system to understand that the person shopping for sneakers is male or female, you need to tell it.

You can’t be ready for anything without machine learning

That means you need plenty of people working on your site, constantly teaching, tuning and updating your search engine. They’ll work hard to makes sure your site is ready for anything. But your site will never be ready for anything. Trying to keep up with a constant stream of more and new products, and the constant evolution of the ways consumers describe them, is a Sisyphean job.

Site search webinar

Screen Shot 2017-03-06 at 8.55.56 PM

Staples is No. 5 on Internet Retailer’s list of the top 500 digital commerce companies. It recently moved from a rules-based search engine to BloomReach’s Commerce Search, a continuously learning search engine that constantly improves its performance based on user behavior and web-wide data. Through automation, the engine is able to achieve one-to-one personalization, while allowing humans to influence search results when they see the need.

Executives at Staples say their site search team now spends one-fifth the time it used to spend manually tuning its multiple digital sites.

By breaking free of tools that rely heavily on manual tuning, retailers are able to let the machines optimize the vast majority of their products and categories while they focus their human expertise on the highest value products, categories and pages.

Free up your human creativity

Why not set the humans loose to pursue tactics and strategies inspired by an analytics system that provides instantly actionable insights? That is, provided your site search system generates actionable insights.

Even Mark Hurd, CEO of Oracle, the company that brings you Endeca, says that maintaining legacy systems costs companies way too much. Citing research that found that more than 80 percent of Oracle’s customers’ spending goes to maintain existing systems, Hurd told CNBC last week that companies should look to the cloud.

“They can reapply that budget and that manpower and that talent to things that can help them gain marketshare, to help the customer deliver better service to their customers,” Hurd told CNBC’s Jon Fortt.

Obviously, the need to hire an army of workers to manually tune a rules-based site is a big expense. But set aside that expense for a minute and consider the opportunity cost that comes with having your best people spending the bulk of their time teaching your site search engine what more advanced systems can learn on their own.

“The higher level of automation, obviously, the more time you’ll save,” RealDecoy’s Richard Isaac said during a webinar sponsored by BloomReach late last year. “A high level of automation simply means the automation can handle most of your heavy lifting, but merchandisers can still customize it.”

The dramatic reduction in time Staples needed to devote to manually tuning its sites, for instance, left its team of retail professionals the time to work on more creative and more high-value initiatives.

The on-premise nature of a legacy system like Endeca can also eat up the time of high-value employees in ways beyond the need for constant tuning. Consider a global brand with a dozen or more different commerce sites all running on on-premise systems.

Any modification or upgrade to the site search engine needs to be installed one-by-one on 10, 12, 20 or more sites. With a cloud-based system, the changes can be pushed all at once, with all of an enterprise’s sites being simultaneously upgraded.

Those upgrades alone are the kind of on-premise projects that require a significant standing IT army or at least an ongoing contract with a partner that can handle the IT needs of a retailer scrambling to keep up with technology and consumer habits that are moving ahead at warp speed.

It’s hard to overstate the importance of site search and obviously every retailer is trying to get it right. High-quality site-search can be a significant differentiator in a highly competitive industry.

Site search is key to digital commerce success

In its work on site search, RealDecoy quoted eConsultancy as saying that 30 percent of e-commerce customers use site search. And when site search is optimized and works well, RealDecoy went on to say, it converts at a rate that is 2 ½ to 5 times higher, citing eConsultancy and Internet Retailer.

On the flip side, offering consumers poor site search is a recipe for blown opportunities. Ninety percent of shoppers don’t go past the first search results page, RealDecoy says, meaning the results need to be relevant and personalized to resonate with consumers.

Isaac put the stakes in perspective — and in dollars and cents — in his December webinar. Say a site attracts a million new annual visitors, he said. Thirty percent of them will use the search box the first time they visit a site. Eighty percent of those will bounce because of a bad search experience. That means you’re losing 240,000 visitors because of bad search.

Now, multiply that by an average order value of $158 and an average conversion rate of 5 percent. By Isaac’s math, you’re out nearly $2 million and that doesn’t count whatever you spent to attract those lost customers in the first place.

That hardly sounds like a way to turn your site search into a revenue-generating engine. If you’d like to hear how to get the most out of site search, join us for our March 30 webinar,  “Three site search moves helping your competition sleep at night.”

Car photo by Ron Reiring published under Creative Commons license.

Mike Cassidy is BloomReach storyteller. Contact him at mike.cassidy@bloomreach.com; follow him on Twitter at @mikecassidy.