This article originally appeared on the BeyeNETWORK
Analyzing clickstreams has migrated from a one-time-only initiative to an ongoing set of processes at many companies. Indeed, many brick-and-mortar retailers have invested large sums of money to simply monitor consumer movement in their stores. They hire specialists in customer behavior, installing cameras to actually track how shoppers get from the furniture polish to the tuna fish, or from one end-cap sale item to another. Experts then pore through hours of videotape to monitor the various pathways shoppers use to navigate around the retail space. One study noted that when consumers enter a store, they normally veer to the right rather than to the left, and that there is a high correlation between left-veering consumers signing their credit card slips with their left hands. Very interesting. But so what?
With any type of data analysis, simply obtaining and analyzing information isn’t enough. The analysis should invoke some sort of business action. Just because I know how someone walks around my store doesn’t mean I know why. And just because I can now track how someone’s navigation of my Web site doesn’t mean I can use this information to incite changes to my business. Clickstream data can indicate a host of customer behaviors. It can tell an outdoor gear e-tailer, for example, where on its site customers spend the most time. But are they there because they’re captivated by the anoraks, or confused about how to find the crampons they’re looking for? Analysis of clickstreams can also indicate that 27% percent of the customers on a child care site stopped shopping once they hit a particular page. But is that because there are no interesting products, or because the baby just woke from her nap?
Deploying clickstreams on a data warehouse to track how customers are navigating your Web site is a good first step, but it’s also just a means to an end. And it’s the end that’s addling companies who have embraced e-commerce and its associated technologies. Now that you have the data, data warehouse, hardware and Internet infrastructure in place to support clickstream analysis, what’s next? And where will it lead?
One of our clients, a general merchandise retailer who has joined the e-tailing ranks, wants its Web site to be as “sticky” as possible and has begun analyzing clickstream data to surmise why customers might leave the site prematurely. The company has honed in on the value of abandoned shopping carts. When a customer leaves the site in the midst of a shopping trip, whatever the reason, the company looks to see what products were in the cart. The data is then compared with similar data from other abandoned carts to examine:
- How much revenue was represented by the abandoned carts? In other words, how much money did we lose by this customer leaving early?
- Whether the products in the cart were high-profit items or loss leaders.
- If the same products were found in other abandoned carts.
- How large were the carts that were deserted, including a rolling calculation of the average number of products in an abandoned cart?
- Whether the total bill for the abandoned carts consistently fell within a certain dollar range.
The result of this analysis can trigger some interesting theories. For instance, none of the products in the cart was appealing enough to a particular customer to motivate him to continue shopping. Or the customer was put off by frequent inquiries asking him whether he was ready to check out. Or that, at a particular dollar total, the customer thought better of the entire shopping trip and bailed. Or that a number or mix of products in a cart reminded the customer of another site that might offer a steeper discount for similar purchases.
Admittedly, some of these theories are mere guesses. After all, maybe the customer’s Internet connection was on the fritz, or the site had a bug that abruptly booted him off. But the fact is, whatever the customer’s reason for leaving the site and a cart “full” of merchandise, the e-tailer can take a variety of sane actions based on its less-than-certain extrapolations.
For one, the vendor can tweak its site design to allow the customer to see a rolling total bill as items are added to the cart, thereby allowing the customer to see the total charge without having to calculate it as he shops, and to check out once he hits his magic budget. Or, rather than requiring the customer to go to another page for specific product information, the site could invite the customer to see pop-up product information with a click of the right mouse button, thereby keeping him in buy mode. Or, it could integrate the clickstream data with more specific customer behavior information, including information from the customer relationship management (CRM) system.
This latter point is perhaps the most intriguing: rather than simply examining a customer’s navigation patterns and guessing about which actions to take, the retailer can combine those patterns with more specific customer data to provide a holistic view of that customer’s value. In certain cases, a one-time-only shopper might have been lost; but, in other cases, a high-value customer might have left the site on multiple occasions. A tailored e-mail message or electronic coupon – perhaps targeting one of the products left behind on a prior trip – could make all the difference the next time that customer decides to log on.
If you already have detailed customer data on your data warehouse, you’re definitely ahead of the curve. Clickstream data in a vacuum won’t solve many problems or foster much improvement. But combining clickstreams with other key data, such as customer value or propensity-to-buy scores, product profitability, or demographic and preference data, could give you a serious leg-up on your competitors.
And retailers aren’t the only companies that can benefit from clickstream analysis. An Internet service provider (ISP) I know has begun using clickstream analysis to better manage its network utilization, tracking its customers’ clickstreams in order to see which Web sites are accessed most frequently. The results have allowed the ISP to cache the top 100 Web sites for a given customer group, enabling those sites to be accessed in memory, thereby saving bandwidth and recouping costs in the bargain. The resulting bandwidth savings have enabled the company to declare solid ROI figures on its data warehouse that weren’t easily calculated with other analysis activities.
Imagine you combined clickstream data with other customer data, either directly onto your data warehouse or on a separate “Webhouse” whose data was integrated with your customer data on a regular basis. Now, imagine you could use data mining to compare the clickstream patterns of high-value customers in order to predict what they would do next.
For example, using preference data on its data warehouse, a financial services firm knows that its best customers – those who have a high average balance, multiple profitable products and do most of their banking online – tend to use the Web site’s so-called Platinum Page, which was created especially for high-value customers. The bank has discovered that in test markets, these customers – by and large – want to integrate financial holdings from the bank’s competitors into a consolidated online “single financial view.” Data mining can help the bank determine whether or not to charge a fee for this service (and how much), as well as which customers would be most likely to pay. If such a service is indeed proved valuable, the bank will not only be first to market, but will have enhanced the satisfaction ratings of its best customers.
Indeed, clickstream analysis for a variety of customer segments, or for certain individual customers, can trigger a salvo of new business actions from more refined target marketing campaigns to revved-up CRM functionality to the development and customization of specific products or product packages.
Exciting stuff? You betcha. But the standard admonishment to walk before you run is as true as ever with clickstreams. If you’re still in the throes of data warehouse acquisition or are consolidating multiple data marts, getting clean, integrated customer, product and revenue data online will give you much more bang for your buck than committing to clickstreams straightaway.
Clickstream analysis should definitely be considered a downstream activity, one that relies on an existing and well-planned data warehouse infrastructure. But once you get there, the water’s great – so jump in!