Web Analytics for Retailers, Part 3

There’s an art to Web analytics. What do you want to know? If you’re not sure, don’t just ask some IT guy to set up your reports or run them in “default” mode (as so many do). Hopefully you’ve had time to read about the most basic and intermediate reports retailers use Web analytics to review. Now, a look at some of the more advanced reports needed to optimize conversion rates and maximize online sales.

Merchandising Reports

Review your Web analytics reports regarding which brands, categories, and products sold best, and study up- and cross-sell reports. You’ll identify which products are most popular and which are likely to be sold together. By identifying top items sold together, you can more effectively price and place products to increase average order values.

Multichannel Reports

Retailers must understand how their customers behave in all channels. Those who don’t adopt a multichannel customer view risk making uninformed decisions. They aren’t taking the whole picture into consideration. The ability to import and export online customer profiles to create a single view of the customer is critical to advanced Web analytics.

First-Time Vs. Repeat Buyers by Products Purchased

Michael Crowdes, manager of interactive marketing and e-commerce at Dirt Devil, uses this WebTrends report for additional insight into what point consumers engage with the company’s Web site.

“More purchases of new models from our Web site come from people who have already purchased something from us in the past. We sell relatively few models online, so it makes sense that most new buyers come to us for replacement items for a purchase they made at a retail store,” said Crowdes. “This shopping behavior seems to indicate that once a consumer has made an initial purchase from our site to support a product they already own, they’re more open to the idea of expanding that behavior to include purchasing completely new products.”

Segmentation Analysis

Every online business has many segments of daily visitors. Some know exactly what they want, some know approximately what they want, others are just browsing or repeat customers. As Eric Peterson, a Jupiter Research (a Jupitermedia Corp. division) analyst says in his book, “Web Analytics Demystified,” “It is a mistake to lump all visitor value groups together and use that information to try and understand how to best serve the groups providing higher value — thus the need for visitor segmentation tools.”

The most advanced Web analytics are performed via profile-based segmentation that uses historical records of online customer activity. Customer profiles allow clients to target marketing efforts based on displayed behaviors. Targeted marketing increases customer retention and lifetime customer value on a daily basis. To do so, employ targeted email campaigns, invest in site tools high-value customer segments use, or correlate online behavior with offline systems to deliver recommendation engines in the call center.

CompUSA segments with Coremetrics’ Web analytics platform. “Our relationship with Coremetrics has improved CompUSA’s online business performance immensely because we have access to a powerful customer analysis platform which helps us identify and optimize for high-value customers,” said Al Hurlebaus, director of e-commerce. “Using Online Analytics 2004, we learned that Product Compare users represented a key customer group for us, sparking an initiative to improve access to that function as well as its capabilities. Coremetrics helped us increase annualized revenues by $2.2 million.”

Learn how different segments behave on your site. Do they view different pages or products? Do they convert at different rates? Do they use different site functionalities? Are the order values different? These are just some of the questions you should be asking.

A/B Testing

Most people end up conducting A/B testing for a simple comparison between two related groups or segments. Instead, they should focus on a true experiment with a hypothesis and evaluation criteria defined before tests are run.

Mayer Gniwisch, vice president of Ice.com said, “Visual Sciences allows us not only to segment a portion of our traffic that will be tested but it also give us the insights to set up proper experimental design and the analysis tools to tell how long the test must run to show statistical validity.”

The tool should allow you to measure actions (conversions) to visitors and calculate based on profits that require the integration of costs, returns, cancellations, discounts, and actual shipping costs.

Keep in mind a few important considerations when considering any of these reports. Accurate tracking of individual users is critical. If IP addresses were used to track individual users instead of unique cookie values, orders may be over-allocated to just a handful of users coming in from large ISP proxy servers. Also, as in our case, the longer the overall lifetime of the product, the more important it is to have a significant amount of historical data to include in reports.

The right reports are critical to achieving a high return on investment (ROI) from your store. Are you getting them?

Related reading