Guessing At What Your User Wants Doesn’t Work

Web-based applications seem determined to embrace a form of automation that really doesn’t make a lot of sense and it’s causing more than a little woe for end users. Personalization, the act of guessing what a user wants based on previous activity, is a serious problem. At one level, personalization is actually costing vendors sales. For example, because of my previous searches on Amazon, I couldn’t find a product I really wanted. Amazon kept suggesting items I didn’t want or need. I ended up buying the product on another site and only later found I could purchase it on Amazon for less money. (Not to pick on Amazon too much, I have the same sorts of problems with other sites such as Google.)

Adding personalization to an application is supposed to help a user find items of interest, but we often need to search for something other than the usual item or may not even know what we want. In these situations, personalization actually gets in the way. A recent article, “Personalization is collapsing the Web” expresses the idea clearly by pointing out that people are sometimes ill-informed about current issues because personalization hides relevant news from view. The problem is so prevalent that it now has a name, the filter bubble, as described in “Content Personalization: How Much Is Too Much?” Users can’t find the information they need and vendors are losing sales due to the filter bubble created by various search engine vendors such as Microsoft (in Bing).

In order to provide personalization, sites must track your every move. It isn’t just the National Security Agency (NSA) that snoops on you, but just about everyone else on the Web as well. To some extent, any suggestion that you have any privacy on the Web is simply unrealistic. However, there comes a point at which the snooping, categorization, and misinformed suggestions get to the point of ridiculous and savvy users start searching for alternatives. One such alternative espoused by writers such as John Dvorak is DuckDuckGo. At least it doesn’t track your every move or provide you with suggestions that you don’t need. The prying can even take odd turns, such as the woman who was arrested for researching a pressure cooker on Google.

Guessing, even if the guessing is informed by previous activity, usually won’t work for an application. In fact, it’s probably a bad practice for a lot of reasons. Here are the things that you should consider when thinking about adding personalization to your application.

 

  • Guesses are usually wrong because few personalization engines can actually understand user needs successfully.
  • Personalization can cost an organization money by hiding products, services, or information that a user really needs.
  • Wrongful suggestions reduce the confidence the user has in the application.
  • Tracking the user makes the user feel spied upon.
  • Time and resources required to track a user and offer personalized suggestions could be better spent by making the application faster, more reliable, and more secure.


Odd as it might seen, I have never yet encountered a situation where personalization was an aid to anything other than making me leave the site for a location that isn’t personalized. When viewing a news site, it’s better to see the top stories of the day and drill down into the less sensational stories as needed. In most cases, what I really need is a good search engine that helps me find what I want, once I know what it is. Let me know your thoughts about personalization at John@JohnMuellerBooks.com.