A Treasure Trove of Data

Like a sports car intended for street use that can nevertheless reach speeds of 160 mph, a new generation of web analytics is now technologically capable of delivering more analytic power than retailers know what to do with, going deep into customer behavior. By contrast, log files, an earlier and still commonly-used method of collecting and analyzing customer activity on web sites, generally can capture traffic to various areas of the site, but not follow a customer’s movements within a site, or more importantly, their meaning.

“Now that we have all this information, the next piece of the puzzle is how to leverage these reports to make changes to optimize the site,” says Jupiter Research Inc. analyst Matthew Berk.

Reviving a debate

Helping to pile up that mountain of data are the increasing number of web analytics solutions available in an ASP model. With user-friendly formats that enable marketers and merchandisers to pull data rather than having to rely on IT staff to do so, and a relatively low cost of entry, ASP solutions are putting sophisticated analytics within the reach of retailers unwilling to buy enterprise software to do the job.

The initial cost of ASP analytics may be as low as $15,000 versus software that can be upwards of $100,000, according to Jupiter.

The ASP vs. software debate is a replay of the old lease vs. buy argument. For an ASP service, retailers pay quarterly or monthly after the initial set up based on traffic volume. Over time, they could wind up paying roughly the same in ongoing expenses as they would for software they could book as a capital expenditure. But ASP solutions also offer time- and resource-saving benefits by delivering reports that don’t require software maintenance and that are immediately useable.

“When you have a small department and limited tech resources it’s nice to have someone take care of that for you,” says Nathan Miller, e-commerce manger at Northern Tool & Equipment Co., which uses Marketforce, an ASP analytics tool from Coremetrics Inc. Retailers like Miller are helping to drive up the ASP side of the analytics market at a rapid pace. Jupiter Research estimates that by 2006, ASPs will account for 29% of all analytics spending, up from 10% this year.

Asking new questions

The new analytics tools are a response to new demands from Internet retailers for the type of customer behavior measurements long available in the brick-and-mortar realm. Aberdeen Group analyst Guy Creese says much of that demand is the fallout from dot-com failures that hauled freestanding and freewheeling Internet retail operations back into the old-line retailers from which they’d spun out. “The minute that occurred, retailers started asking the kinds of questions they’d asked for years about customer segmentation, for example, that were new to this arena,” Creese says.

But the point when retailers were seeking those measurements was also the point when the dot-com investment market dried up and the economy turned. Retailers have struggled to prioritize technology investments since then and many have turned to ASPs. Analytics, new to the industry and with little entrenched base, offered a ready opportunity for many to try out the ASP model.

“CFOs got into the act and saw they could spend a quarter of a million to get a web analytics solution or they could look to an ASP solution with a service level agreement they could cancel if they weren’t getting the results they wanted,” Creese says. “After being seen as a ‘lite’ solution a few years ago, ASPs have gotten better at offering customized services.”

The scientific method

Adopting an ASP approach has made a lot of information available very quickly to retailers because the companies running the ASPs know what they are doing; there is no ramp-up time for an internal technician to learn the technology. And that’s also made it possible for retailers to amass piles of reports which still produce no gain if they don’t yield intelligence that leads to action.

One of the most effective ways to do that is to put the brakes on broadscale “because we can” measurements and start small. Retailers have only to look at how scientific inquiry is conducted to find a model that works in the application of web analytics.

“Some of the most successful retailers approach their use of analytics by thinking about it in terms of an experimental method,” Berk says. “They have a hypothesis, they ground it in data, and then they make one or two modifications to the site to measure the change.”

Take Dartek.com, the online arm of discount computer cataloger Dartek Inc. The 6-year-old web site ventured beyond log files into more sophisticated web site analytics only six months ago when it signed on for HitBox, an analytic tool offered as an ASP by WebSideStory Inc. Dartek, which serves medium and small businesses, home office and individual consumers, started in April with HitBox and this summer became a beta user for the company’s recently released Commerce analytics product.

Separating web visitors into new and existing customers, Dartek found that while the conversion rate among customers of record was on target, the rate among first-time buyers who’d gone as far as starting a shopping cart lagged. “New customers were coming, then bailing out,” says JoAnn McNeely, marketing projects manager for Dartek.

Dartek applied analytics to test theories about the reason for the higher bailout rate. It first checked to see if the problem was one of trust among new customers by featuring a new customer assurance program on the site. But analytics revealed that the change produced only a 1% lift in conversions among new customers.

The company next tested free shipping as an incentive, seeing a 4% to 6% increase in completed sales among new customers who started a shopping cart. But that test also showed up one of the limitations of analytics—they can identify the problem but not necessarily figure out how to solve it. The results of the experiment are so new that Dartek hasn’t calculated yet whether the boost in conversions is worth the expense of free shipping.

Apart from the issue of free shipping, however, Dartek broke even on its initial investment within the first six months, McNeely estimates. Dartek pays WebSideStory a quarterly fee based on traffic. “We are beginning a concerted push from catalog prospecting to online prospecting. We wanted to make sure first that we could track what was successful and what failed so we could properly direct our dollars and effort,” McNeely adds. “We didn’t look at software because it was too cost-prohibitive.”

Merchandising intelligence

Applying the same test-and-measure method to chart the effectiveness of e-mail offers, Skechers.com got a bonus in the bargain: merchandising intelligence that provides a preview of trends to direct future e-mail campaigns, vice president of direct marketing Geric Johnson says.

Top 500 Guide

Shoe manufacturer and retailer Skechers selected 10 products for women and 12 for men for an e-mail promotion to clear inventory on end-of-year styles. Customers received a discount offer applicable within the same session on other merchandise if they opened the e-mail, clicked to the site, and purchased one of the promoted items. The discount is automatically generated by a code in the URL link provided in the e-mail from Skechers.

Analytics tools from WebSideStory showed that the e-mails were being opened at the expected rate, but also provided the information that some 20% more units were being sold in this e-mail blast than with other e-mail campaigns. “Being able to track the performance of a given ad campaign right through to the sale and getting the ROI numbers related to it have been a big help,” Johnson says.

The data already are shaping future campaigns. “The analytics let us pick up on what is hot or what might become hot,” Johnson says. “There are other ways to capture a glimpse of where your business is, but not the relationships of the data. With this tool, I can think out where our business is going to go.”

Searching for reduced spending

At Northern Tool, Miller applied analytics tools to validate click-through rates as tracked by its search engine optimization vendor. That intelligence ultimately allowed Northern Tool to reduce spending on and increase sales from search engine optimization.

“We use Coremetrics’ ASP to measure every online campaign we do,” Miller says. “Any link or banner we have out there has a unique identifier attached to it. We measure click-throughs all the way through the site to sale, and then we use that information to make future campaign decisions.”

As a result of the close measurements, Miller realized that ad dollar spending was higher than gross margin realized from the search campaigns. Investigation revealed that rather than counting each unique visitor as a single click-through, the search engine optimization provider counted any click on any product from the same user in the same session as an additional click-through. “We were paying on a per click-through basis, so sometimes we were paying multiple times for the same customer. The analytics helped us to identify that,” says Miller.

Northern Tool used the data to leverage more bang from its search engine optimization buck. The two companies worked out an agreement in which Northern Tool still pays under the cost-per click model, but the optimization provider upped efforts to increase revenues for Northern Tool with better marketing.

The new measures were implemented this summer and the ratio of search-engine-optimization ad spending to sales dropped by 10% from July to September. That, says Miller, amounts to a 10% increase in revenue. Northern Tool pays Coremetrics quarterly based on the number of page views. “We might entertain the idea of software in the future, but now, we make good enough use of the service that it pays for itself,” Miller adds.

Making the investment

Although the application of detailed analytics to web traffic is a relatively new phenomenon, some retailers recognized the value a few years ago. Unable to find analytics on an outsourced basis or that could give them the highly specific detailed results they wanted, they bought software and modified it to measure what they thought was important. Some say that the returns from the intelligence they’ve gained over the years was worth the investment.

In the case of Djangos.com, a web retailer of music, it sparked an overhaul of the company’s business model. Three-year-old Djangos was set up as a web site aggregator of inventory in stores that Djangos owned. Djangos tapped the stores to forward ordered titles to Djangos headquarters in Portland, Ore. There, it maintained a warehouse with 60 employees to pick, pack and ship orders for customer fulfillment.

But using analytics software acquired from WebTrends—now NetIQ Corp.—Djangos quickly discovered by tracking customers’ ZIP codes 40% were within 15 miles of a store.

Djangos deduced that central shipping was redundant and store ownership an unnecessary expense. As a result, Djangos last year sold its stores. It then signed those stores on as affiliates, and began using its web site exclusively as a storefront that lists inventory in affiliate stores.

Instead of centralized fulfillment, Djangos fulfills orders out of whichever in-stock affiliate is closest to the customer and takes 15% of the sale. Djangos has 40 affiliate partners and is growing that number across the country.

Djangos then focused on building a template for a streamlined shipping system to be used by all the partners. And in July, it launched an extranet that links inventory databases of all the partners to aggregate and display in-stock and out-of stock levels on Djangos.com in real time.

CEO Steve Furst estimates Djangos has spent more than $75,000 on analytics software, with the biggest chunk of that in the initial investment two and a half years ago. Djangos has since written applications internally that build on the original package.

Recently, NetIQ has supplemented its analytic software offering with an ASP model, but Furst says he’s satisfied with what he already has in place. “We haven’t investigated the ASP because we’re quite happy with the tracking we get,” he says.

Determining causality

Whether a retailer uses an ASP model or buys the software, however, one issue remains the same, and that is the difference between compiling data in a report and understanding the report. “Just because something is correlated doesn’t mean you can control causality by duplicating those conditions,” Berk says. “A lot comes down to who is issuing the report and what their training is.”

While that learning curve builds up, some of the simplest applications of analytics, for now, involve using them to identify barriers to the completion of defined tasks on the site. “Removing those barriers to commerce-related tasks is of great value,” Berk says. “We’ve seen retailers who’ve had pretty dramatic increases in their top line as a result.”

mary@verticalwebmedia.com