Web Analytics in Retail: A look at REI

Recreational Equipment, Inc., more commonly known as REI, is a leading outdoor equipment and apparel retailer. Since 2005, REI has used Coremetrics as a web analytics platform to monitor and analyze the browsing and purchasing behavior of visitors two their two online stores (Demery, 2005). In this post I'll take a look REI's implementation of a web analytics platform, look at one specific case where web analytics were used to improve the REI online shopping experience, and offer some further suggestions on how web analytics might be used to further optimize REI's online stores. Before I begin though, I'll provide a little background on the company itself.

About REI
REI is a leading outdoor retailer founded as a cooperative in 1938 by a group of mountaineers from Seattle, WA. Since then REI has maintained its structure as a cooperative while becoming a nationwide retailer of outdoor equipment and apparel with annual sales over $1.5 billion (Society Consulting, n.d.). It counts over 3.5 million people among it's members and is the nation's largest retail coop (REI History, n.d.). Along with stores nationwide, REI has two online storefronts: rei.com and rei-outlet.com (Demery, 2005). Online sales are a major part of REI's business. And the company has consistently been ranked highly in Internet Retailer's Top 500 list, coming in this year at 64 (Internet Retailer, 2012).

REI's Implementation of Coremetrics
REI uses the web analytics Coremetrics to monitor and analyze the browsing and purchase behavior of customers on their website (REI Coremetrics, n.d.). As I mentioned above, REI has been using Coremetrics since 2005 when it was selected because it was a top of the line web analytics platform for retail sites.

In using Coremetrics, REI offers its customers three different tracking levels: anonymous visitor, total opt-out, and cancel opt-out (REI Coremetrics, n.d.).

1. Anonymous Visitor: This is the default level of tracking access, and the one we are most used to with many other web analytics platforms like Google Analytics that use anonymous tracking. With this setting no personally identifiable information is collected (REI Coremetrics, n.d.). 
2. Total Opt-Out: With this setting Coremetrics will not collect any information about a customers "shopping actions" or personally identifiable information (REI Coremetrics, n.d.). This seems to be an opt-out from all tracking, although the language is vague in its specification of "shopping actions." Does this mean that REI and Coremetrics can track actions not related to shopping? 
3. Cancel Opt-Out: The name of this preference level does not accurately describe it. It is actually the most robust level of tracking. With it Coremetrics can conduct the same tracking as they do under the Anonymous Visitor level, but it can also capture the customer's email address and membership number, if the customer is and REI member (REI Coremetrics, n.d.).

While the opt-in page is not prevalent in REI's website, it does set an example of good practice by allowing a customer to set tracking based on their own privacy preferences. If they succeed in getting Cancel Opt-Out permission from an customer, REI has the ability to collect some very robust data about the customer's interaction with the site and even determine information about preferred brands and activities.

Putting the Analytics to Work

Recently, Seattle-based Society Consulting aided REI in solving a specific problem related to variations and unpredictability in page load times that they found their customers were experiencing as their online sales grew (Society Consulting, n.d.). In observing metrics from their website, REI found that lag time in loading pages had a significant negative impact on sales. Research has suggested that the impact can be as large as a 1% drop in sales for an added delay of just one-tenth of a second (Kohavi & Longbottom as cited in Society Consulting, n.d.). 

Knowing this REI charged Society Consulting with the task of determining the effect of page load time on REI customers and then running a cost-benefit analysis on several load time factors, including conversion optimization, customer experience, outage losses, and fault triage (Society Consulting, n.d.). 

Through their project, Society Consulting created a framework for collecting and analyzing performance data built to work with Coremetrics. With this in place, they created a new model to determine the sales impact of changes in load times. With this model, Society Consulting created maximum load times for each section of REI's website and created a dashboard where REI's analysts could easily monitor the site's performance and the impact on conversion rate (Society Consulting, n.d.). 

As a result of Society Consulting's work, REI was able to determine that nearly two full days of sales each year are lost because of slow page load times. In addition, REI was able to more efficiently direct its resources towards developing features and fixing outages that have the greatest impact of sales.

This specific example, shows how REI took one metric (page load time) and found the way it was impacting sales. With the knowledge gathered from their analytics information, REI was able to create a system that could potentially have a very significant impact on future sales. 

Using Analytics to Further Improve Sales

The ways REI could further use web analytics to improve future sales are too many to mention here. While some of my ideas are probably already in uses by REI, I will mention a few. 

1. Use geographic customer data collected by Coremetrics to determine cities for new retail stores.
2. Create funnels from various sales promotion and marketing materials, including emails, catalogs, and coupons.
3. Use data collected from those customers grant full tracking access to tailor individualized marketing messages. This could be emails for new products in the specific activities they participate in, an email highlighting an accessory for a product they recently bought, or sales promotions for the brands they buy most often. 

References

Demery, P. (2005, February 17). REI implements hosted web analytics from Coremetrics. Internet Retailer. Retrieved July 7, 2012 from http://www.internetretailer.com/2005/02/17/rei-implements-hosted-web-analytics-from-coremetrics

Internet Retailer. (2012). Top 500. Retrieved July 7, 2012 from http://www.internetretailer.com/top500/list/

REI History. (n.d.). History & today. Retrieved July 7, 2012 from http://www.rei.com/jobs/rei-story.html

REI Coremetrics. (n.d.). Information collection options. Retrieved July 7, 2012 from http://www.rei.com/shared/help/opt_out.html

Society Consulting. (n.d.). Case study: Web analytics REI. [PDF Document]. Retrieved July 8, 2012 from http://www.societyconsulting.com/case-studies/REI-Case-Study.pdf

Goals, Funnels, and Filters

Tracking simple metrics with Google Analytics is one thing, but to actually start using the tool for analysis that can inform your business and/or communications strategy takes the implementation of more advanced features of Google Analytics. When you dig down deep enough, Google Analytics can provide fairly robust analysis. In order to begin moving you Google Analytics strategy from measurement to analysis some of the first advanced features you should implement are goals, funnels, and filters. In the remainder of this post, I will discuss goals, funnels, and filters individually as I apply them to my own blog.
Goals
One of the most important things you can to maximize the potential of your Google Analytics profile is to set ups goals for your site (Logren, 2012). Setting goals is all about tracking conversions. And as Lars Logren puts it, “Conversions pay the bills” (2012).
Goals are essential to tracking the success of a site. There are four types of goals available in Google Analytics: the URL destination goal, the time on site goal, the pages per visit goal, and the event goal (Google Analytics IQ, 2011). Each of these goals can be used for tracking a specific type of conversion. For this blog, I’ve chosen to set up two types of goals for Adventures in Web Analytics: a URL destination goal and a time on site goal.
I’ll start with the time on site goal since it is relatively straightforward. The time on site goal might be very useful to a site like this blog where the bulk of user interaction comes from spending time on the site reading posts. If a visitor remains on the site for a certain amount of time I can assume that they are interacting with the content I’ve created. With this in mind I could set time on site goal to be reached after a visitor has spent five minutes on the site.
Another goal I’ve chosen to set for this blog is to track when a visitor views an individual post independent of the blog home page. I’ve done this by setting up a URL destination goal with a goal URL that accounts a visit to any post made in 2012 by setting the URL to ^/2012/.* with a regular expression match type (Figure 1). I realize that this will only work for the current year’s posts, but as of right now this seems like the simplest way to capture a view on an individual post beyond the home page.

Figure 1. Setting up a URL Destination goal to track the view of any individual post from 2012.
Funnels The URL destination goal provide a good segue into the concept of funnels. Funnels allow you to define a step-by-step progression through your site that leads up to your specified conversion goal (Google Analytics IQ, 2011). There isn’t a need to setup a funnel on a time on site goal, but a URL destination goal is a perfect place to do this because the destination URL could be a confirmation or thank you page that results from the completion of a sale or an email subscription.
Setting up funnels allows you to see how visitors are progressing through your site, and where exactly they might be exiting the funnel (Google Analytics IQ, 2011). For instance, you might find that 50 percent of your visitors exit your site after step 2 of 6, but that less than 10 percent exit if they make it through step 2. It would stand to reason then that you should find a way to improve step 2 to reduce the exit rate.
On my blog, I’ve set up a simple funnel along with my URL destination goal for a visit to an individual blog post (Figure 1). You can see that the funnel only has two steps. The first is a visit to the homepage and the second a visit to an individual post. This is nothing to complex. However, by setting up a funnel, I’ll be putting myself in a position to better analyze how visitors are interacting with my site and whether it’s a way I’ve intended for them to interact with it our not.
Funnels are much more powerful than I’ve made them seem with this example. The best execution of a funnel will have more steps and a conversion goal with a set value (Google Analytics IQ, 2011). This is something that I haven’t institute on my blog as of yet.
Filters
The final feature of Google Analytics that I’ll explore is filters. Filters allow you to do exactly what the name implies. They allow you as the account owner to set parameters on what data is displayed in a specific profile and how exactly it is displayed (Google Analytics IQ, 2011). With filters properly implemented you can do things like exclude internal traffic to your website, thus eliminating an inflation of metrics by your own employees. You can also change the way data is displayed so that you can more easily read your reports (Google Analytics IQ, 2011).
Before you implement any filters on your site, it is absolutely crucial that you set up a separate profile that maintains the raw data collected from your site. Once you apply a set of filters to one of your profiles, you are unable to go back and recapture that raw data after it’s been processed (Google Analytics IQ, 2011). Lars Logren even recommends setting up one further additional profile to test out new goals and filters without risking the integrity of your main profile(s) (Logren, 2012).
Needless to say, my first step in setting up a filter for my blog was not setting up the filter itself, but actually setting up an additional profile for the raw data. With this second profile in place I proceeded to set up a filter to exclude any traffic from my own IP address (Figure 2).
This simple filter of internal traffic should be the first filter you setup, especially with a blog. It is likely that you will visit your own blog over and over again, thus inflating your metrics. By excluding your own traffic, you can make sure that you get the most accurate portrayal of traffic to your site.

Figure 2. Filter setup to exclude traffic from my local IP address.
References
Google Analytics IQ. (2011). Google analytics IQ lessons. Retrieved June 20, 2012 from http://www.google.com/analytics/iq.html
Lofgren, L. (2012, April 2). The 8 Google Analytics features every site must have enable. Retrieved July 2, 2012 from http://blog.kissmetrics.com/8-google-analytics-features/