First things first: this site is a zealotry-free zone. A lot of people love Google and Google Analytics (GA) specifically – all the more power to ’em. GA is a tremendously useful and powerful product; I’ve used it for many years both personally and professionally and have a lot of respect for the great people that work on it.
But I’ve been concerned about their sometimes curious choices to sunset useful products (Reader, Feedburner). Don’t get me wrong; the products are theirs to make those decisions (and I’m sure there are logical internal business decisions for why they chose to sunset them), but it doesn’t make the choices any less problematic for those products’ users.1 Between this concern, a desire for real-time analytics, and just a general curiosity to see what else was out there, I started looking around for another web analytics solution for my personal use. Hey, it’s also a great opportunity to learn some new tools, right?
Originally, my only real requirements were comparable functionality to Analytics from someone who isn’t Google and whose business model wasn’t predicated on user data. In terms of specific functionality, I wanted geolocation and campaign tracking, as well as being able to easily check in on the go (a native app certainly wasn’t required, but responsive web was the bare minimum).
I initially tried GoSquared and Heap Analytics which, like Google, run professional hosted services. Both are primarily targeted to businesses but offer free tiers for those of us too inconsequential to fit into that target market. In the end however, their feature sets are too specific for their target users that they just didn’t pair up well with my personal needs for this humble little site.
As my research and dabbling continued, I really warmed to the thought of a self-hosted solution. I don’t have anything against hosted services per se, but having complete ownership of my data is compelling. Of course “free” (as-in-beer, but free-as-in-speech too would be better) is a great feature, which dovetailed with the desire for self-hosting.
Way back in the day I remember thinking that if I ever wanted to try something other than GA, I planned to use Shaun Inman’s Mint. Back when Google Reader’s demise was announced, I had bought a license to Inman’s Fever and was impressed with his work. While Mint isn’t free (beer or speech), it’s inexpensive (USD$30 per site), the licensing terms are acceptable to me, and the software must be self-hosted. It’s also extensible, offering both 1st and 3rd party “Pepper” plugins (although some of the extensibility I’m looking for doesn’t appear to be offered by Mint or any Peppers). Unfortunately, much like Fever, Mint doesn’t appear to have been receiving much love lately – they both work, but Inman’s focus and interest are clearly elsewhere. Given this, I couldn’t justify spending the money on Mint (despite being happy with Fever).
Nosing around on Dreamhost’s admin panel, I found that they offered a one-click install for Open Web Analytics – software I wasn’t familiar with, but the thought of offloading the minor maintenance tasks onto my hosting provider certainly made it compelling enough to give it a whirl. Taking a look at their site, it’s easy to see its appeal to someone who’s coming from GA:
Upon further inspection however, the features offered are limited and the extensibility through the module (plugin) architecture is … not really popular. For example, the geolocation module was pretty lackluster especially when compared to Piwik’s; not only did it require manually downloading & installing the city database from Maxmind (which is updated monthly), but there is no option to get visitors’ network provider information.2
While I’m sympathetic to Peter Adams’s schedule and interests aside from being the solitary maintainer of OWA, ultimately I want a tool that’s in active development and gets enhancements from outside parties.
Piwik appears to be where it’s at for self-hosted and free (beer and speech) web analytics solutions. It’s actively being developed by a company that makes its money from premium services on the product; it’s a clear and proven business model that doesn’t leverage my data in order to sell to advertising customers. The feature set is pretty comparable with GA, even if they made some strong decisions in differentiating the UI and experience – it’s not entirely my cup of tea, but it doesn’t get in my way of using the product.3
Unsurprisingly, I decided to go with Piwik.5 There really was no contest here, honestly. Almost by the lone virtue of the dust covering it, OWA was a non-starter. Heap and GoSquared are simply over-designed for my needs.6 If my personal web experiments ever become complex enough to warrant it, I’ll definitely give Heap and GoSquared another look (as well as any players new to the scene) but I’m guessing that between Piwik’s current features & plugin ecosystem and active development I might never need to do that.
That said, it’s extremely hard to imagine Google choosing to do the same to Analytics – especially given how valuable the data they collect through GA is to their actual Search business model. ↩
And even after all of that, I never was able to get the geolocation data to consistently work. ↩
It’s worth noting that I’m not holding up GA UI/UX as a gold standard here either. ↩
I can’t believe I didn’t originally include this – as you might have already inferred given it being a self-hosted solution, there’s no waiting around for your data. ↩
Full disclosure: As I noted earlier, I’m weaning myself off Google products – not cutting cold-turkey. I’m continuing to use GA redundantly with Piwik for the time being until I’m confident in the latter's numbers and performance. ↩
Some follow-up research surfaced Logaholic and that Mixpanel offers a free tier. In the case of the former, it looks interesting – although the massive number of features that only become enabled when you start paying was a little off-putting given that I’m not looking to enroll in a paid subscription for my analytics tool at the moment (maybe later, should my needs change). Regarding the latter, while my site easily qualifies for the sub-25,000 data point ceiling of Mixpanel’s free tier, I’m a little unnerved at the thought of tipping over into their lowest paid tier purely due to having an unusually popular month (plus, I suspect that it might be over-designed for my needs like Heap and GoSquared). ↩