As promised in my last post, Cookies - A History Not A Recipe, today I'm going to reveal my favorite cookies, which are -- Browser Cookies. They're different than the cookies that people eat. They're small files that publishers can use to store little bits of information on the computers of website visitors. I like them because they give publishers a way to recognize browsers that have been to their websites at sometime in the past. They also let publishers store settings, allowing them to remember your preferences when you return to their sites.
We use cookie names as anonymous IDs for browsers. This allows us to track the advertising that people see and do interesting reports for advertisers. For example, we did a report a while ago that showed that prior exposure to display advertising seemed to make people who clicked on search ads more likely to purchase products. Reports like this can help advertisers when planning and evaluating their advertising campaigns.
Some people have concerns about cookies and use programs to delete them on a regular basis. This complicates my work a little, but most browsers seem to have cookies with histories that are long enough to do useful analyses. For example, most ad-serving impressions that we track go to cookies that were created at least 3 months earlier.
The fun part about my job is that there are many very simple but interesting questions that we can answer by looking at our cookie records. For example, how often do browsers get exposed to ads that come from our ad-serving system? Since we serve a lot of ads on a lot of publishers, this question is similar to the questions of how often online ads are seen and how often major publishers are visited. And the answer is surprising: On average, we serve ads to browsers in the US less than 3 days a week.
The graph on the right shows the distribution of US browsers with histories spanning a 13-week period by the number of days that the browsers received ad-serving impressions during the period. The peak on the left indicates that many browsers do not seem to be exposed to advertising very often. I was surprised by this at first, but I guess it seems reasonable. For example, if you have a work computer and a home computer, you might use one of them much more than the other.
Two other things stand out to me in this graph. One is the dip at an average of 5 days / week with ad impressions. My guess is that this dip comes from computers that are used only at work. The small peak at the right also stands out to me. It represents the browsers that were exposed to ads every day in the 13-week period. That's a lot of computer usage, and those browsers must get a lot of ad impressions.
Often, once we get an answer to one of these simple questions, it can lead to many other questions. For example, what percent of ad impressions go to those browsers that are exposed to ads every day? Maybe we can look at that in a future post.
Erik Hanson
More from the Microsoft Advertising Institute