On Fri, Nov 9, 2012 at 11:32 AM, Joe Brockmeier <[email protected]> wrote: > On Fri, Nov 09, 2012 at 09:52:28PM +0530, Rohit Yadav wrote: >> How do other Apache projects set up their site analytics, or do they even do >> that (track user requests)? >> >> The idea is good but I would support the analysis and data used >> to do the same that is public; so analysis no. of issues, bugfixes, >> commits, emails exchanged over ml to get analytics of CloudStack >> as an opensource project; so I really like the *stacks analysis: >> http://www.qyjohn.net/?p=2427 > > The community analysis is nice, but that gives us nothing in terms of > *user* interaction or how successful (or not) the Web site is. > > I'm looking for: > > - how many users visit the site? > - what pages do they visit? > - how long do they stay on those pages? > - where did they come from? > > (Analytics would also tell us about browser, OS, etc. - but really, I > don't think that's terribly useful in our use.) > >> May be just have a simple counter of >> how many visitors visited, as a privacy and eff supporter I don't like >> the idea of tracking user agent data, I'm fine with anonymous counting, >> analysis of visitor data, no. of downloads etc. but that's just me. > > > We'd need an analytics package to get even a simple counter, pretty > sure. We could set up something like a site image stored on S3 or > another site that is called every time someone hits the site, but that > wouldn't tell us much. > > I'm also an EFF supporter, but we're not trying to do anything nefarious > with the data - just see where we can improve the Web site. > -- > Joe Brockmeier > Twitter: @jzb > http://dissociatedpress.net/ >
Given our international audience, I believe that we need to err on the side of caution here. I'd suggest following the lead of many EU focused sites, and be explicit about any tracking cookie. We'd need an opt-out as well. Another option would be to simply use the web server's logs. Infra has some details on how to get access to them. That would mean that we wouldn't need cookies, and we'd lose a bit of data around individual user paths... but we would get the raw page hit counts.
