I’ve always felt that SEO was, to use highly technical terminology, a lot of crapola (which is highly annoying to a business partner of mine who knows a lot about the subject), and not in small part due to the, to use highly technical terminology, random flakiness of web analytics in general. It’s not a science, and I’m not sure it’s an art. But, for better or (probably largely) worse, if you build web sites, you have to deal with it.
For this piece of crap blog, I have some sort of web analytics being recorded in a number of places:
- I use a hideously old version of SubText, and so it tracks web views and agg views. digression: I have had it on my to-do list to try to upgrade to the latest version, but I just have this feeling that the first attempt (and subsequent x attempts, where x > 3) will go hideously wrong, so I keep putting it off.
- : since they are the major player, I have the JavaScript thing on various pages (though apparently not the rss page). However, it looks like Google has released a tool to block some of its own data, so who knows what will happen there).
- : I forget how this is implemented, but for the last year or so, RSS tracking has been through this.
- Clustrmaps: this is the map thingy (technical terminology) on the right nav side. Forgot how it is implemented.
- SmarterStats: once I moved the blog to DiscountAsp.NET, I got this Silverlight enabled thing with a whole heck of a lot of canned reports and the ability to create a host of others.
The main reason why I have all of these things is to see how they work and how they compare to each other. Smart people like Ayende use their blog for marketing purposes, but I’m not that smart, and the target audience for my piece of crap blog is some weird combination of people who might be interested in some combination of: general technical posts, cqrs for dummies’ fans, hockey fans, and fans of well written songs that tend to center around the desolation of looking back at failed relationships (if I could fit in left-handed Albanian albinos, it would be the ultimate in niche web sites, but I digress).
An odd thing about comparing all of these sources of web analytics data is that, at times, it is as if they were tracking different sites. There isn’t always any obvious correlations between them. There are some obvious things I am aware of (like apparently not having the Google Analytics’ JavaScript on my RSS page), but some other things I don’t know what to make of.
Within the last week or so, my Feedburner stats have, to use highly technical terminology, ‘gone all ape-shit.’ I got my Feedburner account a long time ago, but never bothered to implement it for a long period of time until finally thinking it might be a good idea (I think all I did was enable Feedburner syndication in the Subtext options and it did whatever it did automatically). Once I did that, the stats have been fairly consistent on a day to day basis. A slow but steady rise from ‘almost no traffic whatsoever’ to ‘almost no traffic whatsoever, but more than before’ with a few bumps up and down, but mostly pretty stable.
Until the last week or so, where, if you were to trust the numbers, the number of people subscribing has, on one day, dropped 75%, and then the next day, gained it all back, and subsequently all over the place. Since almost all of the subscribers come through Google Feedfetcher, it’s like Google changed how it works or how it records them on a daily basis.
When I look at what SmarterStats is telling me about access to the RSS page, I don’t see any significant correlation. It’s almost like the subscriber number now more closely (but not perfectly) correlates with the reach number, which has always fluctuated wildly. My Google-fu has only come up with this 3-year old page to give some indication, but nothing to suggest there’s been any change to how Feedburner works recently.
Since my blog serves largely to feed my ego and just serve as a dumping place for things I find interesting, I don’t care about the numbers per se, but it does make me wonder how I would handle explaining this sudden fluctuation to a client that was using Feedburner. I really have no idea at this point.