Finding Gold
Powered by data on the back end and sporting some spiffy new visuals in front, today’s site search serves up more than just results
By Mary Wagner
Time was, the site search function on an e-commerce site had one job—deliver a list of available products to shoppers looking for specific items. Sealed in a silo, site search had no connection to other systems or web searches that brought shoppers to the site.
Times are changing, and e-retailers today use site search to do much more. For instance, they can tie the order of search results to business rules aimed at achieving targeted goals. And they can push to the top of search results relevant products with the deepest inventory or those they most need to move as a season ends, a practice dubbed searchandizing.
The latest iteration of site search rolling out on e-commerce sites and from site search vendors addresses the divide between web search and site search with technology that uses data gleaned from shoppers’ web search behavior to serve up more relevant site search results.
The evaporating barrier between how e-retailers use web search and site search to drive sales isn’t the only way the site search of today differs from that of a few years ago. With help from new technologies and the influence of Web 2.0 social exchanges on online shopping, site search is starting to look different, too. Ajax technology that speeds refinement of search results, search clouds that give shoppers an instant, graphical depiction of what’s hot, even search results that serve up video “salespeople” are emerging to grab shoppers’ attention and draw them farther down the path to purchase.
What Ratnaker Lavu, vice president of technology at Macys.com, has to say about what Macy’s has learned about site search and how it’s leveraging that knowledge in new ways might apply to any online retailer looking for more out of site search. “There is more data available today,” he says. “Analyzing that data and coming up with features and functions that improve the experience is an ongoing process.”
The right page
A new blend of site search and web search is boosting conversion at CableOrganizer.com, an online retailer of products for taming the growing hornet’s nests of electrical wires and cables that help keep homes and businesses running. About 70% of CableOrganizer.com’s customers arrive at the site via web search, mostly from Google. But natural search doesn’t always send web searchers to exactly the place on the site that has what they’re looking for, says chief operating officer Paul Holstein.
For example, consumers who did a web search on “label printer” and clicked on CableOrganizer.com’s listing in the results would be delivered to the site’s home page, which doesn’t feature label maker products. “If you look at our home page, it doesn’t seem to have anything to do with label printers. So people were bouncing out,” he says. “For some reason, my home page had higher rankings for ‘label printer’ than the correct landing page did. But I can’t control Google’s organic search—it’s going to send people wherever it’s going to send them.”
While the interface with web search didn’t always work as Holstein wanted, CableOrganizer.com’s site search function from SLI Systems Inc. was effective at getting visitors who used the search box to the right products on the site. Holstein’s question: Why couldn’t web search’s power to expose the site to a huge online audience be combined with SLI’s ability to get searchers to exactly the right spot?
Getting shoppers who reach a site by means of organic web search to the right landing page usually is a job for search optimization rather than site search. But in July, after six months of development, CableOrganizer.com was ready to test something new. It wrote JavaScript that immediately captures the referring domain and keyword of an inbound web searcher. That JavaScript then runs the domain and keyword data through the site’s search.
Next, site search intervenes to send a web searcher to what it, rather than the web search engine algorithm, deems the most relevant landing page on the site. The data transfer that makes it all happen occurs instantly, in the second between when a web searcher clicks on the CableOrganizer.com listing in web search results and when he’s delivered to the site.
CableOrganizer.com has filed for a patent on its innovation and intends to make it available for sale. In the meantime, on its own site, the feature has tripled traffic. Dan Shields, e-commerce initiatives manager, says traffic has risen in part as a function of indexing: SLI indexes the content on CableOrganizer.com as searches on the site accumulate.
Fresh content
Because the SLI function is running more searches on the site, with each click from a web search now counting toward the total in addition to site searches that don’t originate as a web search, SLI is indexing content on CableOrganizer.com more frequently. Search engine spiders read frequent indexing as fresh content, a factor that pushes CableOrganizer.com’s listing higher up in natural search. That, in turn, leads more web searchers to see and click on those listings.
“We are performing three times as many site searches, which means three times as many indexes per cycle, so we are bringing our traffic up and in the process bringing in more shoppers, who are converting at a higher rate,” Shield explains.
Prior to adding the JavaScript linking site search and web search, CableOrganizer.com had been converting at an average of about 2%. Conversion is now 2.5% to 2.75%, and sometimes hits a run of 3%. Linked to web search, site search now is doing about 3,500 searches per day, triple its previous volume.
Macys.com is taking a different approach to getting more out of site search. It is applying data dynamically to how it orders and displays search results, rather than leaving the process entirely to merchandisers or static algorithms.
“Site search is an important piece of our overall search strategy,” Lavu says. “We look at search as a three-legged stool including paid search, natural search and site search. We believe all three are important to the future of understanding what a Macys.com customer is looking for as part of the Macy’s brand experience.”
Lavu describes a site search refinement Macys.com added last year via Mercado Software, its site search provider of the past eight years. The e-commerce site had been using static rules to order the sequence of search results. “Now we are getting more into real-time data that suggests what the sequence should be,” he says.
All in the family
Macys.com applies business rules that order search results dynamically according to the requirements of individual families of businesses within the company. For example, if a shopper searches for Polo Ralph Lauren shirts, a rule can order the sequence of search results based on the rate of conversion of shirts or their freshness in inventory. Site search on another family of products might present results according to product margins or depth in inventory. “It’s all data driven, with a real-time assessment of the data,” Lavu says.
This real-time use of data has helped Macys.com improve conversion. Though he wouldn’t disclose numbers, Lavu says Macys.com is experiencing a higher rate of conversion since it started more dynamically applying business rules to site search. With search now producing a much higher conversion rate than browsing, Macy’s expects to build on that with a new guided navigation approach it’s working on with Mercado, one that blends search and browsing.
It uses Ajax technology, which speeds shoppers’ ability to filter search results by attribute because it does not require a page reload to display a fresh set of results. A search on the site for men’s dress shirts, for example, might produce a hundred or more results, requiring shoppers to page through all of them. The new site search functionality will enable a shopper to filter those hundreds of results immediately by attributes such as brand and color. Layered over each refinement will be whatever business rules have been applied to the display of those products.
“Because we will be relying on the site search technology for browsing, and we can use real data, the results set and the sequencing of those results become very automated, very dynamically,” he says. Lavu expects the search/browse combination to roll out at Macys.com by summer.
That will make site search at Macys.com look a bit different. Other new developments in site search, however, can add dramatically different visual elements. In October TigerGPS.com, for example, dropped its original homemade site search solution and implemented technology from Nextopia Software Corp. In December it was among the first of Nextopia’s retailer customers to add to its baseline site search functionality—which includes features such as the ability to merchandise within search results—a new module called a search cloud.
The search cloud, accessible through a link on the home page, appears on the site as a large field that displays a mass of text—model names and numbers, brand names, product and part numbers, and other terms searched on the site. The larger the size of the text of a term in a search cloud, the greater the number of site searches the term has received. So a search cloud provides a graphical recommendation of what products are popular.
The type size of a term—and whether or not that term is included in the cloud—changes as data on what’s being searched is updated daily or weekly. Nextopia’s technology determines the popularity of search terms by adding keyword searches entered in the site’s search box and clicks on the same terms in the cloud. These cloud clicks are, in effect, searches, as they lead to results pages with product listings.
While the search cloud shows what’s popular, so do customer reviews that appear on the site. The cloud, however, provides something different. “Product reviews are a valuable tool. They’re good once you narrow what you want down to a few choices,” TigerGPS.com president Derek Kleinow says. “But if you’ve never owned a GPS device before and have no idea of what to look for, you are not going to start by reading reviews of hundreds of products.”
A silver lining
With the search cloud in place for a few months, TigerGPS.com has begun compiling data on how many shoppers have engaged the feature and the cloud’s effect on conversion. Even without complete data, Kleinow is happy with the cloudy days. “We think it’s valuable,” he says, “to give our customers tools to help them find the right products.”
Search clouds are one step in the direction of visually enhancing site search results. Technology vendor One on One Ads Inc., though, takes visualizing site search results to an even greater height. It has developed an offering that activates a video of a salesperson who walks out onto a site search results web page; the salesperson speaks to the shopper, delivering a short sales pitch on the product and offering a coupon, which is displayed onscreen for click-through online or printing.
One on One Ads produces the videos for retailers. It is paid by companies that sell products through the retailers. It’s a cost-per-click model in which One on One Ads and the retailer share fees paid by vendors every time someone activates a coupon, which thus produces an advertising revenue stream for the retailer. Or, a retailer can buy One on One Ad’s technology at a fixed cost-per-thousand fee. The company is working with a major retailer and consumer electronics maker, whose names it will not disclose, on deployments of the video salesperson linked to site search results.
Retailers are looking for and getting more out of site search. The tighter connection to web search, the more rigorous use of data on the back end to sequence results, the changing look of site search results pages and other tactics all have a common goal: to shorten a customer’s path between looking for the right product and finding it. Retailers that do not make effective use of site search tools and functionality ultimately will lose out to those that do, some experts say. This is what’s keeping many Internet retailers focused on refining site search.
“We want to understand what the customer is searching for on our web site and on the Internet, and which links they are clicking on,” says Lavu of Macys.com. “With that understanding, we think we can improve the experience they have on our site.”
mary@verticalwebmedia.com
Click Here for the Internet Retailer Guide to Providers of Site Search Products & Services
