Evan Schuman's StorefrontBacktalk

The CIO for billion-dollar clothing retailer Guess Inc. had a challenge. He knew their search engine was delivering terrible results and that he would take many months to replace. What should be done in the meantime?

Even CIOs at billion-dollar retailers have to put their jeans on one upgrade at a time. But Guess Inc. CIO Michael Relich found himself in an especially difficult position when web analytics told him that 60 percent of his E-commerce site’s search results were delivering “not found” responses to prospects.

Guess’ site search was a legacy search and it didn’t have any natural language processing, so typos and searches for reasonable terms?such as “jeans”?often delivered no results because the prospect hadn’t typed in the exact brand name. The company quickly decided to replace its system with an enterprise search system and they started evaluating vendors.

But Relich’s team then had a classic IT problem: What to do with the poorly-performing search for the six months or so it would take to have a new system selected, installed, tested and launched?

If they kept the old system up, customers would get frustrated with bad results and might abandon the site, shop elsewhere and think poorly of Guess. If they removed the search function, customer service personnel and other Guess employees and partners would be deprived of the powerful tool, presuming they knew exactly what they were looking for.

Relich’s decision: Keep the search active on the site, but hide it until it’s working properly.

“Before, our search was just a standard SQL Server on our database,” Relich said. “Unless the customer put in the exact terms in the exactly right way, a whole lot of ‘not founds’ came back. Search is a good thing if you’re doing it right.”

Nirbhay Gupta, the Guess senior E-commerce manager, said his team “hid” the search engine by placing it on the screen “lower bottom, where it was just one of the links.”

How important was the search for customers? Gupta reported that shortly after the search engine was replaced and relaunched in its former prominent placement, purchase conversions increased eightfold.

The system Guess switched to came from Mercado Software, which provided an outsourced hosted option. Bryan Surles, Mercado’s director of sales engineering, said that hiding the search capability for an E-Commerce site is a risky strategy, but having a malfunctioning search isn’t much better.

“Hiding your search box? That’s ridiculous. You’re losing money,” Surles said. “You’re forcing people to use navigation as a strategy. Search is paramount. Customers expect to be able to use search on the site. Otherwise, customers will leave.”

Forrester Research E-Commerce analyst Tamara Mendelsohn said that Guess was in a difficult position where there was likely no ideal move. That said, she questioned whether an E-Commerce search function is considered so essential today that removing it might have been unwise.

“Customers are more likely to forgive a minorly frustrating experience?like typing in ‘jeans’ and not getting any results?than having no (search) at all,” Mendelsohn said. But if the results were as bad as 60 percent “nothing found,” it’s really a no-win situation, she said.

“You’re going to have to choose the lesser of two evils. Customers [that get a lot of "nothing found"] are going to get frustrated and so aggravated with the site that they don’t come back,” she said, adding that Guess probably struck the right balance. “It makes some sense. I would have said, ‘Fix it and fix it now’ but there’s always the question of what to do in the interim.”

Greg Buzek, a retail technology analyst and the president of the IHL Consulting Group, said Guess’ predicament is becoming more common. “You can increase customer service or you can do things that frustrate customer experiences,” Buzek said. “What Guess is doing is they’re trying to eliminate the frustrations.”

Buzek would have counseled removing the search entirely until it worked properly, but he added that it’s difficult to make that decision. “By making search hard to find, in essence, they lowered the customer service or at least the perception of customer service,” he said.

The E-Commerce search engine space is especially complex today, as Google and Yahoo try to become the default E-Commerce engine for many companies.

In Guess’ case, the outsourced Mercado approach also delivered an unexpected bonus: a 60 percent server load reduction because the database now longer needs to crunch search lookup requests. “We now don’t have to go to the database until we have a purchase,” Relich said.

As it happened, the IT department isn’t seeing much immediate noticeable benefit from that 60 percent load reduction because “we had enough capacity to begin with. The database was never constrained.”

Forrester’s Mendelsohn dubbed the server load reduction “an extra silver lining to the whole thing, which is not typically factored into the ROI benefits.”

Relich was more pleased with the scripts in the package, which makes the search engine easier to update, as web analytics identifies more common typos or synonyms that customers are typing. “Before, I had to give it to a programmer to write code. Now all I need is a new rule and I can have a merchandising coordinator do it,” he said.

Guess’ situation is somewhat different than a typical E-Commerce site because it has such a high percentage of multi-channel shoppers. CIO Relich estimates that “85 percent of my site visitors also shop in the store” and “almost 50 percent of our Web visitors come (into a physical store) once a week or more.”

Part of that is because Guess the company is experiencing a radical change in its 25-year history, from an apparel manufacturer that distributed wholesale to a company that today is primarily a retailer that sees 75 percent of its revenue coming from 325 Guess stores in North America (100 in Canada and 225 in the U.S.).

The shift for IT is substantial. “We had a datawarehouse for wholesale, but we’ve never had one for retail,” Relich said, adding that they are now using a package from Microstrategy to create a retail datawarehouse.

Guess is now looking at its Web site as a true sales tool, but one that might do best by helping store sales instead of performing direct sales. The company recently changed its inventory process to try and become more seamlessly multi-channel.

Traditionally, Guess.com was given its share of inventory and was treated as just another store. The problem happened when a particular product was especially popular. The site’s realtime inventory application would remove the item from all pages the instant it detected no remaining inventory allocated to the Web site.

At that moment, the site no longer was a way to investigate all products the company offered. The system removed those items from display even if the warehouse had plenty of that product, but it was not assigned to any store or for the Web site. “We don’t want to offer anything that we can’t ship,” Relich said, adding that when someone then assigned some of that merchandise back to the Web site, “it would suddenly reappear back on the Web. It became a big customer satisfaction issue. We had the item, but it was just on the other side of the distribution center.”

Relich’s team worked with Manhattan Associates to modify the software so that anything sitting in the general warehouse inventory would be available for E-Commerce sales.

Back in the stores, Guess is also now allowing customers in a store to order any merchandise that the store doesn’t have in stock by using a Web interace from within that store, with all shipping charges waived.

This delivered two benefits to Guess: making the customer happier (“The store orders it for you and, three days later, it’s on your porch,” Relich said) and collecting customer E-mail addresses for future CRM options.

Guess has also recently started moving its stores closer to realtime inventory, courtesy of new DSL connections. Before, it was using store standalone POS systems to dial in once a day with updates. The only close-to-realtime option would have been Frame Relay. “It was a very expensive proposition” to get realtime inventory and sales data. “Is it worth a quarter of a million dollars a month to know that?”

But with DSL becoming available in much of the continent, the situation changed. “As it has become more ubiquitious?and for $74 a month?we now have the bandwidth and the store has access to all kinds of information it never had before.”