Search engines that process billions of search queries per month analyse data coming into their logs.
From processing this information you can see what people search at different times of the year, different times of day, what men and women search for, and where the user is located.
All this data plays a fairly important role in determining the most suitable results for a given query — and in choosing the contextual ads shown to the user.
In February, Yahoo patented a method for collecting and analysing this kind of search information. The document describes how the search engine gathers data and decides what information to provide users.
For this purpose, time-based targeting of content (similar to ad campaigns) was used. “Time-based” means it can be based on day of year, season, time of year, dates of upcoming holidays. Past search history is also used in the process.
Yahoo shows contextual ads based on the query and on the content of the site where the ad is placed. But suitable ads aren’t always available. In that case time-based content targeting can be used. For example, analysing queries by time makes it clear that people more often search stock quotes in the morning, and swimsuits and sunglasses in summer.
If Yahoo analyses search queries in real time, the data lets you draw conclusions about which times of day certain keywords become most popular. Aggregating data over several decades could potentially reveal which topics interest users periodically and what users expect to see in results tied to a holiday or season.
This data can be used in many ways, including determining the most relevant ads to display. Several types of user data the algorithm uses include search queries, page views, ad clicks, purchase history, user profiles, and various supplementary information about the popularity of concepts and trends.
Such analysis provides huge amounts of data about what interests people at any moment. It could potentially reveal interesting search patterns and also increase ad click-throughs even when no suitable ad is found based purely on query or content.
