Plugging Wakoopa Into Web Browsers
At Wakoopa, we develop technology to follow the digital lives of our panelists and present that data to our customers in a powerful and beautiful way. Currently, the central functionality of our products is tracking browsing behavior by looking at the websites a panelist visits (of course, with explicit consent). Although we derive complex information such as visitor paths, the central nugget of data that we collect continuously is a website address and the amount of time a panelist spends on the associated site.
We feel it is extremely important to collect this data as transparently as possible, which means that the panelist should almost never need to interact with our software. The technology we have developed in order to achieve this extracts addresses from the location bars of most major browsers in use today: Internet Explorer, Chrome, Firefox, Opera and Safari. As long as a browser is running on the panelist’s computer, our software monitors the addresses and combines them with data such as the current time and whether the panelist is actively using the computer or passively viewing, e.g., a YouTube video.
Last year, however, we implemented advanced features that challenge the unobtrusive nature of our software. From the (market research) industry, the interest in measuring online advertising impact rose considerably. Tools to assist (or even initially enable) this research in a novel way would be a natural extension of our existing solutions. This was an opportunity that we had to act upon. Unfortunately, data about display ads can not be extracted from the browser location bar, so it was necessary to develop an entirely new set of software modules.
After painstaking research, browser extensions proved to be the way to push forward. Chrome, Firefox and Safari can all be extended by writing JavaScript code. The notable exception is Internet Explorer, which is still the most prevalent browser among our panelists (although its usage is steadily declining). Unlike the other browsers, its extension mechanism is rooted in archaic technology from the 1990s, which posed a whole set of challenges on its own. Nothing that could not be overcome by a good investment of time and stroopwafels, however.
With all browsers, there are two major issues when trying to use extensions: (1) how will we keep the extensions (specifically, the code that finds advertisements) up-to-date and (2) how do we minimize the interaction that is required of the panelist when installing the extensions. The former is a technical challenge that we have solved by building a framework for dynamically distributing new extension code to our panelists on the fly. The turnover time for this distribution is very low and is transparent, enabling us to continuously adapt to the changing landscape of online advertising.
Minimizing user interaction during the installation process requires more insight into panelist behavior. Firefox, Safari and Internet Explorer, for example, require users to explicitly accept each new extension. Our tracker tries to maximize the conversion rate for this interaction by guiding the panelists using pop-up windows that explain why the extension is required and urging them to accept the installation. Chrome requires no such user interaction. Careful monitoring of conversion rates proves that these screens are effective in increasing the number of successful extension installations.

Today, our software can monitor website usage as well as detect online display ads. Our software is actively maintained and we dynamically adapt it to changes in browsers and the web. We believe that, with the development of browser extensions that strike a nice balance between functionality and user interaction, we have laid the foundations for many more advanced features and research tools to come in 2012 and beyond. Of course, the basic software requiring no interaction always remains an option for our customers that are interested mainly in gaining insight into website visitation paths.
- Marten Klencke, Desktop Engineer
