How Yahoo Might Determine Buzz

When you have a heavily visited web site, a portal; where people can buy and sell things, perform searches of the web, make selections of topics in a directory, create alerts on different topics, join groups, and perform many other activities, you might be able to tell a lot about the visitors to your site, and their changing interests.

Or, at least you might if your analysis of your log files, your measure of user activity, and your reporting of that activity will allow you to do so. Yahoo was granted a patent Tuesday on a monitoring system that would enable them to categorize those activities, and track the use of different topics and terms used by searchers, or clicked upon.

This kind of buzz is referred to and defined in a number of ways within the patent, including the following:

In one embodiment of a traffic monitor, events are associated with topics or terms and are grouped by category. For example, when a user provides a search server with search terms and then selects a page from search results, the resulting page hit might be associated with one or more of the search terms used. When a user arrives at a particular page after navigating a subject directory, the page hit might be associated with the subject of the navigation. By comparing changes or trends in the traffic associated with a search term or a category, the “buzz” associated with a topic, term or category can be assessed.

What kinds of benefits might be associated with being about to identify a “buzz” about terms or topics or categories? Here are a few identified in the patent:

  • – Letting visitors of the site know what is popular,
  • – Permitting advertisers to focus upon those topics or terms in their campaigns,
  • – Identify cultural trends,
  • – Tracking interest in specific brands, and,
  • – Measuring the effectiveness of marketing campaigns,

The patent is an interesting read, once you get past the “Claims” section. While it’s dangerous to read a granted patent or a patent application, and assume that the assignee is doing what is described in the patent filing, documents like this one may provide some insight into how a search engine might be analyzing and interpreting data derived from how visitors use the pages and services it provides. Here’s the patent:

Web site activity monitoring system with tracking by categories and terms
Invented by Janet Yoo, Kian-Tat Lim, Stanley Ben Wong, and Elliott Yasnokvsky
Assigned to Yahoo
US Patent 7,146,416
Granted December 5, 2006
Filed September 1, 2000


A traffic monitor provides statistics of traffic using an activity input for receiving data related to activity on a server system. Events being monitored are binned by topic or term, where the terms are associated with categories. The categories can be a hierarchy of categories and subcategories, with terms being in one or more categories. The categorized events include page views and search requests and the results might be normalized over a field of events and a result output for outputting results of the normalizer as the statistical analyses of traffic.

The patent was originally filed in 2000, before Yahoo really had much of a search engine, and examples from the document focus more upon the use of an internal site search, and selections of pages from the Yahoo! directory than of a large scale web search engine. But the underlying concept of attempting to understand what users of the site are interested in can be helpful in more ways than those listed in the patent.

What would you do if you had that kind of information at your disposal?

4 thoughts on “How Yahoo Might Determine Buzz”

  1. Good morning

    Do I detect a Rand Fiskin close with a question tactic?

    And I notice your photo up there now – hi 🙂

    Back on topic – you would imagine this is already being done in some fashion? It would be fairly trivial given the data that is collected (SERP clickthroughs)?


  2. Hi Richard,

    I do imagine that this is done already on a larger scale than described in the patent. And if it isn’t, Yahoo needs to do more reorganizing than they presently are. Knowing what the hot topics are at any one time can inform choices of what news to show on the front page of the site, what products to display most prominently, what areas to invest more effort towards. That information from the kind of analytics described should be shared internally in a timely and meaningful manner.

    SERP clickthroughs and analysis of multiple search sessions, mouseovers, toolbar usage, how and when people use different services, and other ways in which people use Yahoo should be among the tools that many business analysts at Yahoo are compiling and using everyday.

    The photo is fairly new – thanks.

    The “close with a question” post approach is one that predates Rand, though Rand should get loads of credit and respect for wielding the approach so effectively. 🙂

  3. Verrry interesting post! If I had the ability to monitor people like that, I would strictly use it for phishing purposes so I could be like those people in the capital one commercial with the funny voices.

    Kidding, this type of info is perfect for dynamic content / ad serviing / email marketing / etc. [soapbox] Marketing on the web is getting so unbelievably laser targeted. [/soapbox]

  4. Hi Chris,

    I was reminded of the character Cayce Pollard in William Gibson’s novel Pattern Recognition, who had the talent of being able to tell what would be a fashion hit, and what would fail to catch peoples’ eyes. The reality is less dramatic, but this still just focuses upon peoples’ interactions with Yahoo.

    Though stuff like a site search can tell you about things that people are looking for, and failing to find. And the other type of information can be used to understand what people are most interested in, what topics are hot, and so on.

    Do you stock extra inventory of Wii’s in your stores, when the Wii becomes the most searched for product on your search engine? I would.

    Do you take your knowledge of the different interests of different demographics, and build services and content for those? Again, probably. I guess the difficult part is in taking this information, and using it as effectively as possible.

Comments are closed.