This is part of the Wiki Wednesday and coincidentally also part of the Web Analytics Wednesday series.
UPDATE: changed throughout, some spots still not filled in, I am promising some graphs; want to work in the RichmondWiki blog notes as well.
UPDATE2: analytics for wiki edit failure
Google Analytics has a bunch of new features. Mediawiki, the code that runs Arborwiki, has Google Analytics installed and is tracking user behavior to help improve the quality and relevance of the content that's there. Here's some applications of the new features.
1. Segmentation by geography.
Arborwiki covers information about Ann Arbor, Ypsilanti, and the surrounding Washtenaw County area. One reasonable segmentation of the readership and contributor base is to see how people from inside the area behave compared to people from outside the area.
...
(graph)
We'll break down usage for specific pages and see how they differ. First, look at traffic coming from Wikipedia - how does it compare to everything else?
(graph)
Next, look at the most popular page, the "birthday deals" page, who is seeing that?
(graph)
A further refinement will look at how much birthday deal traffic is coming from surrounding suburbs, e.g. Oakland County.
2. Conversion tracking: who edits a page.
We have set up the idea that a page edit is a success criterion and a conversion goal. More page edits, the better. Please contribute.
There are at least two end states: a page is edited that didn't exist before, or a person edits a page (because they linked to something that wasn't there) and it was abandoned. That's a failure condition and a sign that the wiki is incomplete in that specific area. So we set up reports to look for pages where people are editing the page but it shows up as the exit page, which implies that someone left the page in mid-edit
The method:
- segment by page includes "action=edit"
- top content report (or perhaps
- show "exit rate"
- the higher the exit rate for edits, the more people bailed out; look to improve anything that's red?
(graph)
(report)
3. ...
...
(quote from Avinash)
(quote from ...)
...
You can overthink a civic wiki; until there's a critical mass of content, report generation is not that much good. So don't stare at your metrics until you have some idea that there's enough there to work from - a map of the holiday light shows (Richmond Wiki), the great places to go for a birthday dinner (Arborwiki), [your favorite civic wiki's favorite page].
----
Web Analytics Wednesday meets worldwide: ...
Wiki Wednesday meets worldwide ...
Other civic wikis: RichmondWiki
