Optimizing Wikipedia's Fundraising Banner Ads(blog.wikimedia.org)
blog.wikimedia.org
Optimizing Wikipedia's Fundraising Banner Ads
http://blog.wikimedia.org/2009/12/11/annual-fundraiser-checking-banner-results/
5 comments
I've found them far more obnoxious over time, personally. As the first commenter notes, this is also a legitimate concern--but there's no easy metric for obnoxiousness. (Clicks on the "hide" button perhaps?)
Yes there is: measure the frequencies each banner is shown to each visitor and add that data to your optimization parameters. Of course this requires you to identify and track each individual visitor, but it's a standard practice in online ad systems.
no easy metric for obnoxiousness
Pick metrics you care about, and look at them instead then. Editing activity, likelihood of first edit, account signup, what have you. (This is on the if-a-tree-in-a-forest-makes-no-sound-who-cares-whether-it-fell-or-not theory.)
Alternatively, ask people questions like "What do you like most about Wiki?" 6 choices. "What do you like least about Wiki?" 6 choices, including ads.
Pick metrics you care about, and look at them instead then. Editing activity, likelihood of first edit, account signup, what have you. (This is on the if-a-tree-in-a-forest-makes-no-sound-who-cares-whether-it-fell-or-not theory.)
Alternatively, ask people questions like "What do you like most about Wiki?" 6 choices. "What do you like least about Wiki?" 6 choices, including ads.
I have written a blog post on this with detailed explanation of the 'anchoring heuristic'.
http://saperduper.org/post/293243288/wikipedia-donators-anch...
I also posted it on HN http://news.ycombinator.com/item?id=1007587
http://saperduper.org/post/293243288/wikipedia-donators-anch...
I also posted it on HN http://news.ycombinator.com/item?id=1007587
This made me think to "Predictably Irrational" by Dan Ariely too, quite a good example of the anchoring phenomenon.
http://www.ted.com/talks/dan_ariely_asks_are_we_in_control_o...
Btw, i agree with those that says that is quite hard to come to any meaninful conclusion just using the average and max donation, why not analyze also the median value or the std dev/complete donations distribution?
I still don't understand why Google doesn't just support Wikipedia. That whole "knowl" thing didn't work out that well... it would almost be a feel-good thing for them. Hell, google.org could do it on behalf of the parent corporation as a charity move.
a) statistical significance?
b) average is probably not as good a measure as median, since I doubt it's normally distributed.
b) average is probably not as good a measure as median, since I doubt it's normally distributed.
Are these banners really better than advertising? (What revenue would half the pixels, devoted to tasteful ads instead of a pledge drive, yield?)
Fundraising drives don't raise questions about the impartiality of the content in the same way that ads might. However, even NPR has sponsors these days (but not ads per se). I think it'd be interesting to hear whether Wikipedia has considered that approach.
I don't know about that. My local PBS station experiences a lycanthropy where they degrade into infomercials for older people every pledge week. They very well know the kinds of programs that pull in more funds.
If the Wikimedia Foundation at any point decides to use their "business sense" and look at the articles that raise the most funds, well... there are questions that will be raised.
If the Wikimedia Foundation at any point decides to use their "business sense" and look at the articles that raise the most funds, well... there are questions that will be raised.
It's a non-profit. They're not interested in maximizing revenues.
Non-profit has nothing to do with revenue. Every non-profit that survives has to consider revenue carefully and improve efficiencies.
Particularly ones that require a lot of infrastructure and maintenance to keep running. (servers)
They are interested in 'maximizing impact'. If less staff effort (and less screen real-estate inconveniencing users) could generate more sustaining revenues, their mission is better served.
Works for Mozilla! (So far...)
Works for Mozilla! (So far...)