Web Personalization Panned by Jupiter Research

Remember the hype surrounding Website personalization?  Articles from 1999 through 2001 asserted that personalized sites were the wave of the future ---- soon all sites would be personalized to display only content that was of interest to the visitor, and they would update dynamically based on the user’s clickstream to display timely offers, tips and suggestions based on what the user had viewed on the site.  The incentive for site owners to provide personalized content was that by doing so they would increase the “stickiness” of their site and thus their sales or conversions. 

Early reports of the return on investment for personalization projects were very positive. In 1999, Precision Marketing reported that a study of the top 25 online retailers found a 300% return on investment for personalization projects and gave specific examples like the Music Boulevard site, which experienced up to 30% customer click-throughs to purchase on personalized offers versus 2 to 4% for non-personalized offers.  Personalization software vendors also claim very positive results.  A case study on the Broadvision site asserts that Northeast Ohio-based Invacare will save $83,500 annually in information requests that will be self-served from Invacare’s website. An American Airlines case study on the Art Technology Group website asserts that ATG’s Dynamo product has been responsible for a 25% increase in traffic to the American site because of improved user experience.

More recent studies of personalization in the press have reached very different conclusions.  An October 2003 study by Jupiter Research of personalization was titled “The Personalization Myth” and concludes that “the ROI of personalization is dramatically unproven everywhere we’ve looked.” The specific issues raised by Jupiter include:

• Insufficient analysis by companies investing in personalization technologies.  38% of companies surveyed by Jupiter said they planned to invest in personalization in 2004, but “of companies that determine online budgets as a percentage of revenue, only one in five used comparable metrics to justify personalization ROI.” 
• Expense.  Jupiter estimates that a personalized site costs more than 4 times as much to operate as a comparable dynamic site.
• Better alternatives. Had companies investing in personalization done research, Jupiter asserts, they would have found that many other improvements --- such as improved navigation and better search capabilities --- would yield better results than personalization for a fraction of the price.

Does all of this mean that personalization as a strategy for driving Web traffic and revenue is dead?  Probably not.  As one personalization advocate stated in response to the Jupiter survey “Anything can be done expensively and badly.  That does not mean that personalization as a technology is not worthwhile.”

The ineffective, high cost scenario-driven personalization solutions reviewed by Jupiter are just one flavor of personalization, and they are the most complex, high effort and expensive flavor at that.  Other solutions, such as integrating a standard content management solution with a CRM system and simply providing content based on user preferences (rather than trying to predict a user’s interests based in their click pattern) is inexpensive to implement and relatively effective.  Another related technique is to tag content with keywords that allow pieces or even sections of content to be mapped to a standard taxonomy.  This taxonomy allows for more effective search and can be employed in conjunction with the CRM-CMS solution described above to good effect.