Facebook has teamed with Microsoft to create Graph Search, and they’re betting it will replace Google as the number one search provider. The premise (nicely articulated by Dave Williams in a recent AdAge post) is that by including what you say you like on Facebook into a search algorithm, searches will be better targeted to your needs. Better results will then yield more interest, ultimately dethroning Google as king of search.

It’s not a bad premise, but it is ultimately a flawed one. The problem is, people don’t share enough of themselves on Facebook. Instead they flit from post to post, sharing interests superficially, without depth. If you want proof, look at the results from targeted ads versus social data. While social data has value, it’s not more valuable than behaviorally targeted ads. There is a hierarchy at work in the data. Let’s call it the hierarchy of intentions.

The Hierarchy of Intentions

A hierarchy is designed to show the most important factors at the top, then descend in order to the least important factors. The hierarchy of intentions is accordingly designed to rank types of data about consumers by it’s ability to project intent.

At the top is purchase data. Buying anything is a great indicator of interest. After all, the buyer was interested enough to spend money on the item.

Second is search, closely matched to word-of-mouth. Because both are organic, both indicate a high level of intent to act. You might argue that word-of-mouth is not data. This is true, except where it occurs in the social sphere. So if semantics and keyword tracking and analysis uncover an interest in any type of item or service, then that can be transformed into data and used to evaluate intent to act.

The next level down finds clicks, opens, category browsing all lumped together as another level of organic interest. While not as strong an indicator as purchase or search, they do show some level of interest.

The lowest level includes reported data – both the demographic information available for purchase from data providers, as well as the self-reported information about likes (for example) on social networks. This information can age quickly and tends to be less reliable than actions. But it has value too.

Facebook’s Graph Search is betting that information from the lowest level of the hierarchy will help make search so much better than what Google offers, that people will see the value and move to Facebook as the top search provider. However, Google – with a huge base of email already used to drive targeted ads – has the ability to use keywords from emails (word of mouth – the second level in the hierarchy) in their search result algorithms. Which should give them an edge over the Graph. So I wouldn’t count Google out yet.

Does Graph Search have potential? Sure. But maybe more for Bing – since the Facebook partner and Hotmail/Outlook owner has the potential to combine data from both 2nd levels and 4th levels of the hierarchy into its algorithms. And that’s something Google just doesn’t have yet. Then again, neither does Facebook.