How many page-views are you really getting? Part 1 – The WordPress Stats Test

Years ago, it used to be that hits were the metric websites looked at to see how popular they are. That’s changed. Now, amongst other things, page-views are king.

A few years ago, I started blogging. Almost immediately, I signed up for a Google Analytics account. I soon discovered that there was an enormous disconnect between what Google Analytics reported vs. my server stats. The reason for this disconnect can be explained by differences in the way they measure traffic.

WordPress, Google Analytics (and others like Quantcast & ShareThis) all capture data through the execution of a script, or page tag or some similar mechanism that runs whenever a visitor loads a page at your blog or website. The problem with this is that an increasing number of users now have script and java blocking enabled on their desktop-based browsers. They also block cookies. Many mobile devices are not captured by these mechanisms for the same reasons. Consider this as well. Most government offices and medium to large sized businesses block cookies and java script by default. Whether people at work should be surfing to your site is an entirely different issue. The reality is; they do!

The server side stats on the other hand capture everything. They miss nothing (well actually they can miss the ‘odd’ thing due to caching).  The problem with the sever stats however is that they report every search engine, spider and bot that hits your site.

What does this disconnect look like?

Here is a snapshot of the Life As A Human traffic for April 22^nd as captured by WordPress Stats and by our sever stats.

22-Apr	Page-Views
WordPress	3,391
Server	40,584

Look at the enormous difference in page-views. Our Google Analytics stats reported similar numbers of page-views as WordPress did. The lion’s share of the difference between the two is due to spiders, bots and also the way that pages are loaded within the WordPress framework. Reality is much closer to what Google and WordPress reports. By the way, Life As A Human  had 184,951 hits that day! Ahhh .. for the good old days!

So how close are WordPress and Google to reality?

I have been experimenting with the use of StumbleUpon Advertising to drive some traffic to the site. I started using StumbleUpon for advertising at synaptici.com many months before we launched Life As A Human. In short, it costs 5 cents per referral to your website or blog. So, for $5.00 you can buy 100 page-views and target a specific demographic/geographic. It’s a great service and companies like Warner Brothers, The Huffington Post and NBC, use it, with much deeper pockets than ours!

I noticed when I started to do this that when the campaign was done, StumbleUpon would report having sent 100 referrals. My WordPress stats however showed only 90 or 95. As mentioned earlier, this is due to users using script blocking. I wasn’t really that worried about a 5 or 10% discrepancy. Recently however, I noticed this delta was growing. I decided it was time to run a simple test to see if I could measure the difference.

The Test

This is a very simple test that any one can run from their blog or website.

• I created a hidden page. I called mine ‘hidden test page’. Pretty original eh!
• I set it to: no index / no follow – so that search engines, spiders and bots could not see it and affect the test results.
• I e-mailed 50 people and asked them to participate in the test by surfing to the page (once) and letting me know via e-mail when they had. I also asked them to let me know if they had surfed to the page on a desktop or a mobile (or both) so I could capture the difference between the 2. Alas, this part of my test failed as I forgot to ensure I was capturing that data on the server side. Dang! However, this did not affect the main results.
• I let the test run for 5 days then took the page off line.

Here are the results

Actual Page-Views	60	% Captured
As Reported By WordPress	45	75%
As Reported By The Server	60	100%

That’s a pretty significant delta, don’t you think? The server data contains no spiders or bots. It lines up ‘exactly’ with the number of e-mail replies to the test.

Ok, so what does that mean to me?

• If you have no intention of monetizing your site, other than the morale boost of realizing that you are getting more page-views than you thought, probably not much. According to this test, if you take your WordPress reported page-views and add 33.33% to them, you will have a more accurate total.




• If you are a small to medium size publisher (website or blog) and you are monetizing through ad placement mechanisms, then, at least according to this test, you ‘might’ be leaving 33.33% of your impression ad revenue on the table. I don’t believe there is much you can do about that but knowledge is always empowering. Certainly, the advertiser won’t be in a hurry to see this change. They‘re getting more bang for the buck! Why change that? If you have some ideas, please, share them in a comment.




• If you are a larger publisher with your own marketing team and you are handling your own ad placements, you probably already know about this and have taken steps to make sure you aren’t leaving that much on the table by using your own, more sophisticated analytics tools. If you haven’t, you really should.

From a bigger picture point of view, I believe what this trend shows is that in the coming years, this issue will most certainly need to be addressed. There are several solutions but none will be perfect. Please understand, I am in no way suggesting that there is something nefarious about this. This is not a conspiracy to keep dollars out of our pockets. It is simply an artifact of current traffic measuring methodologies.

Likewise, this is not a shot at Google Analytics or WordPress Stats! If you don’t have a Google Analytics account, you really should. They have done an amazing job of developing a very powerful tool that can provide you with a wealth of information about your traffic. Not using it is to deprive yourself of data that can help you grow in ways you might not have even imagined. That people and companies block cookies and scripts is not their faults.

Another  Test – Part 2 – Google Analytics

As I mentioned, this was a simple test and it is in no way conclusive. Further and much more robust testing is needed to more accurately quantify the actual delta between the traffic measuring mechanisms mentioned in this article. This is simply a snapshot. I just happen to believe it is indicative of the current realities associated with the measurement of website traffic, and I know I am not alone in this belief.

I will be running another test sometime soon. I will be looking to capture the difference between how mobile devices and desktops are captured. I suspect that we will see a much higher percentage of missed mobile devices. This test will compare Google Analytics data against the server and WordPress stats. Actually, I did include Google Analytics in the first test but the test data yielded +/- 12% inaccuracies so I am not comfortable presenting that data. What I can tell you is that like WordPress, the Google Analytics results were significantly below the actual total.

• A huge thank you to everyone who participated in this test. I hope that you have found the results interesting and more importantly, useful. I also hope you will participate in the second test!




• If you would like to be a part of the test group for the next test, please leave a comment indicating that you would like to participate.




• If you have run these types of tests yourself, please, share your knowledge and let us know.

 

Photo Credits

Thumbnail – Digitage Web 2.0 – Creative Commons – Maureen Flynn-Burhoe

© Gil Namur, 2010