Measuring user behavior on a website can provide strong signals about that site’s quality. For example, if a visitor arrives at a site, visits 10 pages over the course of an hour, selects a product, adds it to her shopping cart, and then buys it, chances are pretty good that she found what she wanted there. Contrast that with the visitor who arrives at a web page and hits the back button of his browser in less than a second.
These are examples of user engagement signals, and search engines are beginning to use this kind of data in their algorithms. The specific signals they are using and how they’re using them is not easily discerned, however. Search engines are secretive about the details of their algorithms because they are important trade secrets, and because it makes a spammer’s job harder. However, we know that user engagement signals.
How Google and Bing Collect Engagement Metrics
Bing and Google both have a large number of data sources available to them. Some of the most important are: Search results User interactions with the search results are a key source of data. For example, if the user conducts a search and chooses not to click on the first or second result but does click on the third, that can be a signal that the third result may be the best result for that query, especially if this is a common occurrence.
Search engines accumulate data like this in high volumes every day. Browsers Microsoft’s Internet Explorer and Google’s Chrome both have sizable market share and could be a rich data source for search engines (see “March 2014 Market Share” and “IE11 passes IE10 in market share, Firefox slips a bit, and Chrome gains back share”). Browsers are powerful data sources because they can monitor every action a user takes.
Potential User Engagement Signals
Search engines can measure the amount of time spent on a given page using their browsers or toolbars. This is also referred to as dwell time, and more time on page might be considered signal of higher quality. Time on site Similarly, time on site could be considered a positive signal if the average user spends more time on your site than she does on the sites of your competitors. Of course, it could also mean that your site is difficult to navigate or loads very slowly, so you’d need to look at this signal in conjunction with other signal.
Another set of signals that search engines can use involves voting mechanisms. These are methods by which users directly indicate their approval or disapproval of content. Here are some examples: Chrome Blocklist extension On February 14, 2011, Google released a Chrome extension that allowed users to block specific websites from their search results.
This was not used as a ranking factor in the initial release of Google’s Panda algorithm on February 24, 2011, but it was included as part of the second release of Panda on April 11, 2011.
Document analysis is a somewhat different concept than user engagement, but it can be used to predict how users will perceive the quality of content on a site. Strong signals indicating a likelihood of poor-quality content could potentially be used as a ranking signal by the search engines.