Change is one of the few things in life that is constant. It’s an old cliche, but especially appropriate when dealing with people who use the web.
Consumers cruise from one web site to another, reading articles of interest, registering for newsletters, and making purchases along the way. At the same time, the sites and ad servers are adding to their profile databases trying to learn more about their visitors.
But what happens to profiles when people change interests? How do the profiles get updated – or do they?
When e-commerce sites were first using recommendation engines, it was common to hear people complain that recommendations were based on having bought gifts for someone else, not on purchases they made for themselves. Many of these basic problems have been solved by sites asking buyers if the purchase is a gift or for themselves. Some sites are taking this a step further by adding a feature so consumers can enter the name of each person on their shopping list.
But what about the profiles of the customers themselves? Why do these profiles get out of whack when it comes to what people really want?
There are several reasons why profiles become out of date, but usually the problem is time – time away from the site, a time of change in people’s lives, or maybe it’s a plain ‘ol time to market issue.
If you sell consumer products via the web, then you know that once you sell a large item that consumers need, they won’t be buying that item again very soon.
Does the average household need more than one digital camera? Usually not. So once you’ve sold the first digital camera to a consumer, your recommendation engine should know not to waste time trying to get them to buy another one. Instead, it should change to suggesting they buy more memory modules – and more film – because the consumer’s profile has changed to needing consumables.
You’ve probably heard a salesperson ask a prospect, “Are you in the market for a product like ours?” The salesperson is trying to determine the profile of the potential customer to know what products to recommend and the likelihood of purchase. The same holds true on an e-commerce site where a customer’s purchase history should be reflected in his or her profile in order to make the best possible recommendations.
Having made a particular purchase is not the only thing that causes a person’s profile to change. Interests change from time to time, and that leads to regular customers becoming former customers. If a customer who used to buy hiking equipment from you is now interested in camping or boating equipment, your site needs to quickly recognize this change and adjust their profiles to keep in step.
Profiles can even change when a product is never purchased. In a business-to-business setting, projects are analyzed, studied, and evaluated. Potential vendor’s sites are visited, and web site profiles are created to recommend the right product.
When a project is delayed or canceled, the vendor web sites don’t know that the window of opportunity to sell to that prospect has closed. Later, when the member of the evaluation team returns to a vendor’s site on a different mission the interest profile is out of date, so there must be a way for the prospect to update his or her interest profile.
These are examples of changes in interests and needs that occur over an extended period of time, such as weeks or months. But there are other times when top of mind interests change more rapidly, such as over a few hours.
This change in mindset is frequently related to the type of content someone is seeing at a particular moment. Many marketers feel that placing ads near content that’s similar to the product promoted increases the likelihood of a response. A profile database isn’t needed to place ads near compatible content, just the fact that someone is reading articles on a particular topic indicates the person is receptive to marketing messages on that topic.
So how do we update the many different types of profile data?
Most sites obtain profile data by observing behavior on the site, tracking purchase behavior, asking questions with forms, or all three. While some web marketers feel one method of gathering profile data is better than another, the reality is that the accuracy of each method varies for different products for a number of reasons.
When possible, collect profile data from multiple sources, then use analytical techniques to determine the value of each type of data in making projections about customer purchases.
One of the best ways to ensure that profiles collected with web forms are accurate is to use that data to personalize content in an email newsletter. We’ve seen many audiences use the profile update feature to change their topic interests so they receive articles relevant to their interests and needs.
But it’s not just the e-commerce recommendation engines where profiles can become outdated. The companies serving targeted ads based on profiles also have the potential for accumulating irrelevant or outdated profile data. The more sophisticated companies use a complex combination of age, frequency, content, and other factors to deliver the best possible targeting.
While it’s hard to learn the details about each company’s methodology for maintaining accurate profiles, asking a few questions about how profiles are updated will give you a good idea of whether they are attuned to this need.
Like outdated web sites, old profiles can create poor word of mouth when the updating technique doesn’t keep up with your customers’ needs and changing interests. Whether you invite customers to return to your site to participate in managing their profiles, or explain how their activity on your site automatically updates their profiles, keeping customers in the loop can help reduce the problems of outdated profiles.