August 16, 2010
Last week, I decided to finally upgrade my blog to the latest version of WordPress. I always try to avoid jumping in too soon on these things. That way, I avoid many of the little bugs and compatibility issues that tend to plague initial releases.
There is value to be gained by being an early adopter, sure. But there is also overhead, and free time is scarce enough without adding anything to it.
The Big Bad Blog has been doing “well”, I suppose, this summer. July was our first month with over 10,000 pageviews which did not feature the help of the Stay Puft Marshmallow Man, and the average number of daily views in August has been nearly triple that same number in March.
Here at the Big Bad Blog, we take the philosophy that these things do not matter — we write for us, dear reader, not for you. However, those who know the individual behind the intrepid blogger are likely unsurprised that he obsessively checks out the site stats. And, it turns out, when one obsessively checks such stats, one wishes for the counts to increase.
Thursday evening, we switched over to the new WordPress. You will see that on the right side of the following graph, just before the dramatic fall:

So what happened? Did the blog go down due to a bad upgrade? Did it need to be re-indexed by Google, causing search results to dry out? Was the brilliant SEO ruined?
The answers: No, no, and ha!
When I switch over to Google, and take a look at their analytics, I see the following:

The same thing is measured, but now the results over the weekend stay at that same high level (and at about triple the level from March), having only a slight drop-off, rather than a dramatic one.
So no, the site did not suddenly break or become harder to find (or — heaven forbid! — become less popular). Something changed in the way WordPress’s native stats application counts visits.
The numbers provided by Google Analytics have, historically, been considerably lower than those provided by WordPress. Google does a much better job of breaking down the numbers, but I have been hesitant to switch over for one simple reason: for the first six months of the Big Bad Blog, we did not use Google Analytics.
Because the numbers offered by the two differed strongly, I felt I could not change. How could I figure out if a new blog post was as popular as the Marshmallow March? The only way is to compare numbers that have been gathered in the same fashion.
The popularity of a website, of a blog post, of a search term, is relative. If I count in one way, and you count in a different way, we cannot compare our numbers. That your number is bigger means little if you counted some things twice and I counted everything once.
Now WordPress has done themselves in. From August 13, 2010 onwards, their own statistics are being provided on a different basis. They are now much closer to those being provided by Google — which would lead us to believing that they are more accurate. Unfortunately, their newfound accuracy makes them less useful — they have changed their apples to oranges, and I now cannot use their stats to compare the old with the new.
Website statistics are only useful when used for comparison — beyond that they are merely navel gazing — and comparing two things requires consistency, not accuracy. I do not care if my results are all exaggerated by 30% (or underestimated by 30%), so long as they are consistently so. Then I know if more people are reading, or less, or the same.
As a result we are moving to Google for our measurements, as they seem to have been using the same measure since we first signed up to their site. And while that might reduce July to a 5,000 view month, at least I know that when August comes in at over 6,000 pageviews I’ve had a 20% improvement.
Whatever that means.