As much as I complain about the fundraising team's continued neglect of the as yet unmeasured potential of the majority of the volunteer-submitted banner messaging ideas, I thought it would be nice to show the results of their latest experiment:
http://i.imgur.com/gDBvNSt.png Note that the experiment is apparently being continued. Details are promised soon at https://meta.wikimedia.org/wiki/Fundraising_2013 The most recent answer to my question about why the remainder of the volunteer banner idea submissions haven't been tested was because the best banners were performing so much better than all the others. But what does that say, statistically? It means that the variance of the population of possible ideas is large enough that there is reason to believe that testing the rest of them would result in further very large improvements. I've been told several times, going back years, that yes, the Fundraising team was working on the ability to do multivariate testing (which would simply involve showing several hundred different text messages in the existing banner framework, and keeping track of the relative performance of each.) Now that we are displaying thousands of banners per day in this ongoing experiment, does anyone know whether there are still plans to do multivariate testing of the volunteer banner message submissions? _______________________________________________ Wikimedia-l mailing list [email protected] Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, <mailto:[email protected]?subject=unsubscribe>
