Hello and welcome,
When not writing the highly entertaining Sean Yeager adventures series, I do what I can to spread the word. A phenomenon I’ve come across recently is a massive disparity between website page read stats from different sources. And by massive I mean 1000% inaccuracies.
So what do we mean by a page read? Simply – how many times did someone open (and hopefully read) a page of your content? (Or more precisely – the number of discrete page impression per browser session / IP combinations by a human being in a given time frame).
(Definitions – IP meaning IP address which can identify a single machine or at least a service provider’s bundle of active connections to the internet. Session being a period of active use of a given website before a log-off / period of inactivity. )
Now that may seem like a very straightforward concept. And it is. However, consider this:
- A website site counter showing 2000 reads with peaks of 100+ per day for a rolling month
- Google Analytics stats reporting 50 reads with peaks of 4 per day for a rolling month
And that’s for the same website and for the same month.
And on Scribd for related content:
- Scribd document counter showing 3000+ reads for a given period
- Scribd user map counter showing 12 users who have ever read the content for the same content and period
So what’s going on here and how can we interpret these conflicting web stats?
Firstly, they are all strictly incorrect as a measure of real people, because that is impossible to measure unless everyone logs in and/or saves cookies for the whole month, while using the same device / log-in. The reason being that no website can recognise and verify the person browsing the content unless they choose to positively identify themselves. (Consider a shared public or home computer or multiple log-in Ids per person or multiple devices per person).
In addition, we have to eliminate web crawling software, malware and robots that trawl for content from our numbers. This should be easy given that their sessions on any given page will be very short and their IPs repeat. However, many counters do not look at such information. Which means that …..
We can only gauge ‘real people’ responses based on comments, subscriptions, purchases and log-ins. Otherwise we are left with vague ‘ball parks’ and trend indicators and that is all.
From my experience Google Analytics under-counts and unfiltered website counters over-count. (I base my Google assertion on tests that bypass Google search results and use JScript =Off with cache clearing each time).
Incidentally, there are no apparent patterns that I’ve noticed to correlate between the two sets of figures either. One might expect a busy day to show up as spikes using either measuring method, but in my experience they don’t. I’ve had a spike on Google and nothing unusual on the website stats and vice-versa. I often have spikes on the website counter and next to nothing reported on Google. Which is frustrating and perplexing.
Bottom line – if you really need to know your traffic stats, you need to find a website SEO tool that is impartial and well integrated with your site / blog. Unfortunately, you can not rely on free tools and counters if you need accuracy. And you’ll need a webserver log analytics analyser to see what is really happening – but that is for the big commercial sites only. For the rest of us we simply need to make do with the free or cheap alternatives.
Hope you found that useful. That’s all for now.
Author of Sean Yeager books