Enterprises' level of insight into customer behavior has mushroomed as the capture of customer data has evolved from snapshots to streams.

In the past, due to the difficulty and expense of continually monitoring buyer preferences and behavior, companies relied on periodic snapshots: that is, administering in-person surveys, convening focus groups, taking telephone polls, following consumers around the store and so on. This meant that they typically surveyed only a sample of the population and often asked people about their behavior, rather than observing it.

As transactions have become more digital, enterprises no longer have to make do with intermittent "slices of insight." Instead, leading-edge companies are now monitoring "streams of insight," continually monitoring the actual behavior of large numbers of customers, ultimately gaining more accurate and up-to-date views of their customers' behavior. There are at least five types of streams: clickstreams, searchstreams, decisionstreams, walkstreams and salestreams. Given how your company tracks its customers, you may come up with more.


The first example is clickstreams - that is, the logs that track visitor movement on a Web site. These logs tell companies how long a visitor was at the site, where they came from, which paths they took, what pages they lingered on and which ones they zoomed through.

Besides helping enterprises optimize the design of their Web site, clickstream analysis has enabled companies to improve other communications - product catalogs, for example. Through online A/B testing of product images, copy and placement, companies have discovered how to best present their products to their catalog customers and - not surprisingly - sales have increased.


Somewhat similar to clickstreams are searchstreams - logs of successful and unsuccessful search queries. A visitor using a search term is on a mission; rather than aimlessly browsing, they are looking for something specific. Therefore, search terms are solid, explicit indicators of what customers are looking for.

Companies can use this data to correct enterprise/customer vocabulary disconnects. For example, perhaps the customer is looking for something they have - a "laptop" - but the enterprise calls it a "notebook" and so the customer never finds it. By accommodating the synonyms that customers use, companies can improve search results and customer satisfaction. In addition, searchstreams are a good way to track customers' evolving interests and "hot buttons" over time.


Decisionstreams are byproduct of online buyer's guides. Operationally, these guides facilitate product selection based on customer preferences; behind the scenes, they log users' preferences and decisions.

The resulting stream of data is useful for identifying emerging customer desires: for example, last month customers were willing to pay $50 for that feature; today, they will now pay only $35. Decisionstreams also let enterprises understand customer trade-off strategies: for example, is the customer willing to pay a 10 percent premium for the Sony brand name over Zenith? Finally, decisionstreams help product managers and brand managers decide on the appropriate feature mix by letting them see which groups of features customers prefer.


Walkstreams are derived from video cameras mounted in the ceilings of stores. But rather than preserving the image of the customer, as in a surveillance video, the software turns the person into an anonymous token that is tracked. Depending on where the cameras are installed, retailers can count the browsers outside the store, track how many window shoppers come into the store and monitor popular browsing paths within the store.

Walkstreams can document store traffic baselines, making it clear that although two stores may have the same revenues, one is much better at converting browsers into buyers. Walkstreams can also suggest reasons for these differences - for example, helping management discern whether increased sales are due to better window displays, better product placement within the store or both.


Another word for salestreams is point-of-sale (POS) data; however, it's not just data from cash registers - it can also include sales information from other sales mechanisms, such as vending machines.

When the information base is rich enough - as it often is in grocery stores that offer loyalty cards - retailers can start to understand the implicit customer missions. For example, was this the weekly "big buy" purchase, or the "topping off" purchase of milk, bread and eggs? When linked to special coupons for a specific customer, retailers can start to document brand loyalty: "Hmm, this customer switched paper towel brands only when he received a coupon for $1 off - in contrast to his peers who were willing to switch when they got .50 off."

Leverage the Streams

Compared to intermittent snapshots, continual streams offer many benefits: 1) they have less sample bias, 2) typically monitor actions rather than intent, and 3) offer continual feedback, enabling companies to view shifting trends as they occur, not months or quarters after the fact.

Consequently, if you are using streams, don't hoard their insight. For example, rather than keeping clickstream, searchstream and decisionstream information within the Web analytics group, pass along actionable findings to product managers. Volume analytics can be powerful, but only if the entire corporation enjoys its benefits.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access