Thanks for the mention Seddon & Anna, tl;dr version: This is a test to see if readers are more likely to donate if the banner is more relevant to what they're reading. Thanks for testing this community-suggested fundraising idea - sign of a new era regardless of the results of the test!
As some may recall, in July last year a "fundraising ideas" page was created by WMF fundraising to be able to give practical suggestions/ideas for improvement: https://meta.wikimedia.org/wiki/Fundraising/2015-16_Fundraising_ideas
I have been vocally about Fundraising's practices and adherence to the 'fundraising principles', so, when a request for specific suggestions was made, it seemed only fair that I should try to submit productive ideas not just critiques! :-) After all - whether we love them or hate them, the fundraising banners are the single most important revenue-raising tool we have and it is no secret that it is becoming increasingly difficult to reach the donation targets [for a variety of reasons].
One of these suggestions on that page was "related content banners" and I'm super-happy that WMF-Fundraising have taken the time to test the concept. Not just because I hope the idea is successful, but also because it indicates in a very practical way that they're trying to build a new era of more collaborative practices. So, even if the test doesn't result in anything useful, I'd like to say "thanks!" for giving a community-made suggestion a go.
Wikimedia fundraising banners have always been about trying to optimise for the most efficient design/text to be used across a whole language edition. We have targeted by country, and by language, and that's it. Most of the other kinds of personal-targeting that online advertising does is not available to us because of our principles and policies (privacy, cookies...). However, what we have never tried to do before is to target the banners based on the vast array of areas-of-interest that people come to read about. If someone is reading an article a sport, a chemical element, a tv-show, a historic battle, etc. then we can assume they are *interested *in that topic. Furthermore, we have systems like Categories and thematic 'wikiprojects' to group these topics (as well as, in this test case, simple manual selection) so we should leverage that!
The hypothesis that this banner-test is investigating is: Are Wikipedia readers are more likely to click on the donation banner (and actually donate) if the message that is displayed is more relevant to the article-topic where it appeared?
As some people have asked already, this does have some potential trademark-infringement questions when it is associated to an article about a commercial product (which is why the legal department was indeed definitely consulted for this test case). But, the targeting doesn't need to be for that kind of thing. It could equally be, for example: - a profile of a prolific Wikipedia feature article writer of articles about Plants displayed above articles in category:botany. - A winning picture from WikiLovesEarth or WikiLovesMonuments), used in articles about places located in that country. - A personal message by a nobel-prize winning chemist, appearing above articles in category:chemistry. - an 'easter egg' banner (somewhat like this Game of Thrones test) where the text is a reference to that specific subject. I can imagine we could do that with articles about Hitchhikers Guide to the Galaxy?) - A/B test the effectiveness of 'normal' banners based on the kind of article they appear on e.g. maybe text-centric banners are more effective on science-related articles, and image-centric banners are more effective on biographies? Who knows... [There would need to be many practical things worked out: avoiding situations equivalent of when the 'Jimmy banner' appeared above the article on Scopophobia (fear of being stared at)!]
An obvious downside to this concept is that the more targeted the banners become, the fewer people who see them. Therefore the cost benefit of making them becomes worse. Put another way - it's more financially useful to improve the 'global' banner by 1% than it is to improve a very-targeted banner by 30%. However, If the main hypothesis proves true, it could be an excellent way for the global community to suggest culturally-appropriate and interesting/clever banner suggestions, and also a way to re-ignite the general public's interest in actually *reading *the banners. Imagine, instead of making the banners more desperate/aggressive each year (which makes the audience increasingly immune to them - known as 'banner fatigue'), we made them *interesting to read*!
Sincerely, -Liam / Wittylama
wittylama.com Peace, love & metadata