August 14, 2003, 12:00 AM

Three tests at takes the conjecture out of metrics

Tower Records ran tests in three areas of to determine how well it was doing in converting browsers to buyers and in serving customers. “Science beats conjecture every time," a Tower exec told the eTail 2003 East conference.

There’s much talk in the online retail industry about metrics. And many online retailers believe that the more metrics you have, the better you can operate your business. But there’s such a thing as metrics overload, Kevin Ertell, senior vice president, direct to consumer, of Tower Records, told the eTail 2003 East conference this week. The problem, said, is sometimes there is so much information and so little time to analyze it. “And too much staff conjecture about what works,” he said.

Ertell ran tests in three areas of to determine how well the site was doing in converting browsers to buyers and in serving customers. “Science beats conjecture every time," he said.

Using Fireclick Inc.’s NetFlame product, Ertell took a look at’s home page. It was cluttered with promotions, limited time offers, specials--up to 15 at some times. The problem was that a low percent of traffic was going to some of the site’s best promotions. Free shipping attracted only 0.3% of traffic. The main promotion got 0.4%. And 65% of visitors went straight to the search box “We were missing an important opportunity for sales with our promotions,” Ertell says. “The site was simply too cluttered. It looked like a Jackson Pollock painting. It was just too much.”

TowerRecords then reduced the elements to a few major promotions, reduced the image weight to make images smaller and easier to load. It moved a major promotion to the center of the page, taking 40% of the home page. It left some promotions on the left and right panels. The result: the promotion garnered 9.4% of the traffic, conversion rates went up 25%, page load times decreased by 38%, Ertell said.

Next, managers turned to the checkout page, using NetFlame to understand how customers navigated the page and Webhancer to collect data from actual users, both at and at competitors’ sites. The goal: reduce abandonment and increase conversions. For a month, Webhancer collected data from its panelists to create a comparison of 15,000 users at Tower Records and two competitors. One area of concern was the registration page. One explained why registration was important and its exit rate was respectably low 5.7%. Another competitor had little explanation and its exit rate was 11.6%. Tower Records fell in the middle at 8%, Ertell said. Dwell time on the three sites was 46.27 seconds at the site with the explanation, 36.14 seconds at the other site and 29.98 at Tower Records. Conversion rates showed equal disparity. 3.88% at the first site, 2.26% at the second site and 2.5% at Tower.

At checkout, things were equally grim. “Tower was the poorest performer of the three sites. We sucked,” Ertell said. Tower Records reduced number of checkout pages from five to three and removed all images from checkout pages so they would load faster. It also increased the explanation of the value of registration. Results: in one week, conversion rates went up 18% and abandonment rates went down 12%, Ertell reported.

Finally, it looked at customer satisfaction using ForeSee Results Inc. and the American Customer Satisfaction Index. Again, the goal was to increase conversion and reduce abandonment. “The difference between other measurements and customer satisfaction is that customer satisfaction measurements help you predict the future,” Ertell said. “If a customer is satisfied, it carries on into their future behavior. No other metric helps you to understand customers and what they need. Highly satisfied customers buy more, are more profitable and are more loyal.”

The ACSI compiles a score for a retailer based on customer levels of satisfaction with a site’s content, functionality, look and feel, navigation, order process, privacy, product browsing, product information, site performance and transaction costs. While Tower Records earned a score that, Ertell’s words, “rivaled Amazon’s,” which is a very high 88 out of 100, the process did uncover problems with customers who came to the site but didn’t buy. They gave Tower Records lower scores on important drivers, primarily product information and functionality. They also complained about the five pages for checkout. They were asked about their primary purpose for coming to the site. “25% of these pre-purhcasers came with the intent to purchase but the conversion rate was nowhere near that percentage and that meant we had a huge opportunity for improvement,” Ertell said. The data are new enough that TowerRecords hasn’t yet determined how will approach the problem, he said.

“A complement of tools is required to optimize customer satisfaction, loyalty and retention,” Ertell said, “but you should start with customer satisfaction. Focusing on customer satisfaction takes the guesswork out of managing a web site. It’s the best strategic tool for managing our most important asset--our customer. I’m really sold on it.”


comments powered by Disqus




From The IR Blog


Cynthia Price / E-Commerce

4 tips for improving email marketing results

Every piece of data you collect can help you serve your audience exactly what they ...


Bart Mroz / E-Commerce

How smaller retailers can utilize data as effectively as Amazon

Smaller companies have more constraints, but once they set priorities can still benefit greatly from ...

Research Guides