Establishing Site Trust and SERPs with User Behavior

The evolution of search will focus on extensively on user behavior. Google, Yahoo, and Microsoft will move towards analyzing and trusting user behavior over other ranking factors because this factor alone is the best measure of a sites worth.

Consider this scenario:

Your task is to create an index of sites ranked by how much a user of your index will enjoy the results. You have a team of the best engineers and behavioral scientists in the world.

Imagine that the only data you have for creating this index is the surfing behavior of 10% of all web users. You will know what they search for, what they click, how long they spend on a page and what sites they visit and revisit.

I would submit that with the proper analysis of this data alone, one could devise a way to deliver more relevant and more enjoyable results than any of the search engines currently provide..

These would be some of the factors you would want to consider:

1. Profiling What is the user interested in? Based on which sites a surfer frequents and on query profiling, we can determine what a interests a user.
a. What sites does that user visit regularly?
b. What are the topics of those sites? (based on prior search queries)

2. Expertise Is the user a niche connoisseur or is his interest more passive? We might say that someone is a connoisseur of a topic based on what sites they visit and how long they spend on sites devoted to a given topic. Someone with a passive interest in Baseball might check one story from yahoo news per day while a connoisseur will be in baseball forums, go to online gambling sites, read detailed analysis about games, check lineups, and read many more articles about a topic.

3. Time investment How long does the user spend on the sites that he visits? If a user follows a search query to a site and then immediately leaves, that would indicate a poor quality site.

4. Do users visit multiple pages on a site? More page would be an indication that the user is interested in what the site has to offer.

5. Bookmarking What percentage of users bookmark the site or visit a social bookmaking site immediately after visiting the site? If a user wants to save a reference to a site, that is an indication of quality.

6. Repeat Visitors What percentage of users return to the site?

7. Usefulness Does the user continue searching after clicking a result? (was the question answered?)

So, for example, we determine that Mike is a Linux geek (connoisseur). Let’s say Mike
1. Makes a specific query related to Linux
2. Clicks on 4 results
3. Spends less than 30 seconds on the first 3 results
4. Spends more time on the 4th result
5. Bookmarks the 4th result
5. Browses more pages on the 4th result site.
6. Returns to that site at a later time.

We could say that Mike found the answer to his query and that the 4th result should be given more trust in future rankings. Over time we could develop metrics or filters to determine how much to trust a given behavior or behavioral pattern – mostly by trial and error, but eventually we would have it.

Google, Yahoo and Microsoft are already incorporating this data into their search algorithms. They just need to work out the kinks and get past the trial and error phase. Analysis of user behavior through monitoring and profiling will be one of the most important Search developments in the years to come.

So what does this mean for webmasters like you? It means that sites that have loyal users and traffic will get more traffic. Do everything you can to increase stickyness, user loyalty, site visits and page views; it will increase the trust value of your site with the search engines. The rich will get richer.

For Black Hat SEOs, this may signal a shift towards zombie networks to emulate users and fake user behavior – not for the purpose of clicking ads, but for increasing the perceived value of a site in the eyes of a search engine’s algorithm. It will be the equivalent of controlling a percentage of the Neilson Boxes. Without sufficient information on what the user behavior really is, this may prove too difficult a task in the long run. However, the early implementations of behavioral analysis in Search will create exploits that could yield incredible returns.

In the future, however, this Zombie Boosting technique will only work with sites that can pass human editorial review – even if only cursory. I will discuss why in the next week.

Stay tuned . . .

Both comments and pings are currently closed.

6 Responses to “Establishing Site Trust and SERPs with User Behavior”

  1. rokit says:

    How will they track my behaviour?
    Cookies are not reliable.

  2. QuadsZilla says:

    Toolbar + computer footprint works.

  3. littleman says:

    Yeah, I think this is the way of the future, and will get to be the dominant factor in rank in time.

    Re: how
    Like stated, toolbar, and user data (referer/IP info + time intervals) but also proxy service could be used to do a representative sample (such as google accelerator). Another thing is advertisement (adsense, YPN) that will give tons of info, user retention, which pages were viewed within the sites and for how long, etc. And ultimately, Google Analytics will just hand G all user behavior data.

  4. Much agreement. The same issues that the engines have delivering relevance you’ve enumerated above are the ones sites have and should be dealing with. UX is holistic in the mind of the user. If they have trouble finding a pair of new pumas they don’t care if it’s the SERP or the link or the site, all they know is that they couldn’t find what they were looking for online.

    Yahoo especially has made a concerted effort in this regard and I see it as they way they are looking to compete with Google in search. It’s no accident earlier in the year Yahoo hired Andrei Broder (http://research.yahoo.com/~azbroder) probably the leading researcher on searcher behavior in the world.

  5. [...] Ho appena finito di leggere un interessantissimo articolo, in inglese, di seoblackhat sull’user behaviour e come questo elemento sarà fondamentale, nel futuro, per studiare e migliore il posizionamento di un documento sui motori di ricerca. A ben pensare, difatti, quale miglior strumento potrebbero avere i motori di ricerca per offrire una pagina di ricerca il più possiibile soddisfacente per gli utenti? Semplice e chiaro: utilizzare gli stessi utenti come parametro di valutazione. [...]