Community Wishlist Survey 2017/Archive/Better readership statistics
Better readership statistics
Outside the scope of Community Tech. The good news is per below, the Analytics team has some plans to implement similar statistics in the future.
- Problem: There's only one pageview statistics tool (that I know of) which only tells the number of pageviews per day for a given article; it is a simple tool that does not differentiate readers from viewers. I think there is a need for more and better data, such as (1) whether an article is simply clicked on but not scrolled (ie not read much) or read with scrolling (2) how many minutes on average an article is read (3) which subsections get read (4) where the eyeball traffic comes from (for example, whether from other Wikipedia articles, or from the web) (5) whether the pageviews were unique readers or the same readers reading multiple times.
- Who would benefit: The community will benefit if we can learn more about how we're serving readers. Better data will help us focus on specific articles, knowing which to improve and so forth. If there was an indicator of how many unique readers that a specific contributor has -- over time -- then it could encourage contributors to earn more readers by quantifying their contributions.
- Proposed solution: I am not technical enough to offer specific solutions but there are talented programmers here who can.
- More comments: Note: if there is such data already available but I just don't know about it, I apologize in advance, but to my knowledge, it doesn't yet exist.
- Phabricator tickets:
- Proposer: Tomwsulcer (talk) 15:50, 9 November 2017 (UTC)
Community discussion[edit]
- I don't think wikipedia uses Google analytics :) good idea, but we could end up revealing more than we want people to see abut our actual readership. A Den Jentyl Ettien Avel Dysklyver (talk) 16:34, 9 November 2017 (UTC)
- I helped author Pageviews Analysis, which I think is what you're talking about. I will try to answer your concerns but note that I am not on the Analytics team, nor can I give any official word regarding privacy policy.With pageviews, referral information (where traffic comes from) is being stored, along with some other data, but this is all intentionally considered private and is not exposed to the public. I agree more detailed statistics could offer some great benefits, but the general notion is that privacy will be compromised, and that is given priority. I can think of some examples, but I will leave that to the privacy experts. Similarly, data on unique devices are not publicly available, except on a per-project basis. For that, check out Siteviews.At any rate, changing the availability of these statistics is not a job for Community Tech. If you would like to learn more, you could try reaching out to the Analytics team directly. They will be able to give you better answers than me, but again I am doubtful such private information will become publicly available, and it should be noted special requests for the data are rarely granted.Given this is not in our purview or within our control, I'm going to have to archive this proposal. We very much appreciate your participation in the survey, and I am sorry we cannot help any further! MusikAnimal (WMF) (talk) 01:36, 10 November 2017 (UTC)
- Well I suppose thank you is in order for at least responding to my request. I have been a significant contributor to Wikipedia, since 2009, writing and revamping hundreds of articles, and with my contributions of images (over 1000), a metric of how well I am doing here, the only metric that I have, is pageviews, and my guess is that they add, over time, to over 200 million pageviews. That is, the idea that people are reading my contributions motivates me and I bet it motivates other contributors too. I continue to think that my request for improved readership statistics is valid, that they would indeed help the encyclopedia by motivating contributors, as well as providing clues about where best to focus our energies (eg substandard wiki-articles with significant readership). My sense is that it is surely possible to provide better readership statistics while preserving privacy, such as by revealing only aggregate numbers or averages, and not reporting numbers if they fall below a certain threshold, say 30 views. So I believe that tabling my proposal is premature and I suggest that you reinstall it.--Tomwsulcer (talk) 10:28, 10 November 2017 (UTC)--Tomwsulcer.--Tomwsulcer (talk) 10:36, 10 November 2017 (UTC)
- I agree! I just don't think Community Tech can do anything about it. It might be nice to still leave this proposal up during the voting phase, just to see how many people are on board with the idea. I will discuss this with my team. We have a lot of time (until November 27) so do not worry, your proposal is not losing momentum here in the archives :) I'll get back to you soon! MusikAnimal (WMF) (talk) 20:26, 10 November 2017 (UTC)
- Ok, thanks again for responding, and if it is or becomes feasible to somehow get better readership statistics, while protecting privacy using aggregate numbers, I think we'd all be better off.--Tomwsulcer (talk) 14:26, 13 November 2017 (UTC)
- Hi Tomwsulcer, I'm Dan Andreescu and I work for the Wikimedia Analytics team. We do have plans to work on some of the statistics you request here, and the results of that work will come out over the next year and beyond. We can get around some of the privacy problems by tricks that include the ones you mentioned (discard buckets that are too small, separate out statistics so they can't be used together to de-anonymize private information, etc.). We have some bigger commitments that we have to finish up through the spring, but after that I personally intend to focus on projects like WikiCredit. This begins to try and deliver to each user statistics that are relevant and rewarding, hopefully. I would suggest you send your request to our mailing list so everyone can see that community members value this kind of work. Milimetric (WMF) (talk) 18:14, 13 November 2017 (UTC)
- Ok, thanks Dan.--Tomwsulcer (talk) 18:18, 13 November 2017 (UTC)
- Hi Tomwsulcer, I'm Dan Andreescu and I work for the Wikimedia Analytics team. We do have plans to work on some of the statistics you request here, and the results of that work will come out over the next year and beyond. We can get around some of the privacy problems by tricks that include the ones you mentioned (discard buckets that are too small, separate out statistics so they can't be used together to de-anonymize private information, etc.). We have some bigger commitments that we have to finish up through the spring, but after that I personally intend to focus on projects like WikiCredit. This begins to try and deliver to each user statistics that are relevant and rewarding, hopefully. I would suggest you send your request to our mailing list so everyone can see that community members value this kind of work. Milimetric (WMF) (talk) 18:14, 13 November 2017 (UTC)
- Ok, thanks again for responding, and if it is or becomes feasible to somehow get better readership statistics, while protecting privacy using aggregate numbers, I think we'd all be better off.--Tomwsulcer (talk) 14:26, 13 November 2017 (UTC)
- I agree! I just don't think Community Tech can do anything about it. It might be nice to still leave this proposal up during the voting phase, just to see how many people are on board with the idea. I will discuss this with my team. We have a lot of time (until November 27) so do not worry, your proposal is not losing momentum here in the archives :) I'll get back to you soon! MusikAnimal (WMF) (talk) 20:26, 10 November 2017 (UTC)
- Well I suppose thank you is in order for at least responding to my request. I have been a significant contributor to Wikipedia, since 2009, writing and revamping hundreds of articles, and with my contributions of images (over 1000), a metric of how well I am doing here, the only metric that I have, is pageviews, and my guess is that they add, over time, to over 200 million pageviews. That is, the idea that people are reading my contributions motivates me and I bet it motivates other contributors too. I continue to think that my request for improved readership statistics is valid, that they would indeed help the encyclopedia by motivating contributors, as well as providing clues about where best to focus our energies (eg substandard wiki-articles with significant readership). My sense is that it is surely possible to provide better readership statistics while preserving privacy, such as by revealing only aggregate numbers or averages, and not reporting numbers if they fall below a certain threshold, say 30 views. So I believe that tabling my proposal is premature and I suggest that you reinstall it.--Tomwsulcer (talk) 10:28, 10 November 2017 (UTC)--Tomwsulcer.--Tomwsulcer (talk) 10:36, 10 November 2017 (UTC)