Your Website Numbers are Lying to You

Who can you trust on the Web these days? Certainly not your Web anlaytics or tracking softeware. At least not without a grain of salt and the ability to selectively ignore data.

Check out these confusing and contradictory numbers from a few of the sites I maintain. The site vary in rank depending on the source or criteria used. The discrepancies in numbers between Google Analytics (and other JavaScript dependent options) and WebTrends (and other log file dependent software) have been well documented for years. But when you mix in these other rankings it further muddies the water.

It’s very important when talking Web stats with a client that your comparing apples to apples (or Google vs. Google), rather than apples to melons (Google vs. server stat programs). If they’re matching previous server stats to a redesigned site’s Google Analytics numbers – you’ve got some serious explaining to do.

It’s surprising how inaccurate Alexa’s rankings seem to be. Especially given that services like Text Link Ads use those numbers as a primary factor in ad link valuation. For example GDoss.com was ranked higher than e-Lacrosse.com – despite the fact that it get’s only a fraction of the traffic.

Rankings and traffic for the month of February 2010

* I haven’t figured out why yet, but when I redesigned this site about 5 months ago the Google PR dropped from 3 to 0 for the home page. Some interior pages still rank a 3. Very strange.

Why Don’t Google’s Numbers Match Up?

There’s a variety of reasons why Google Analytics (GA) numbers won’t match – a generally be much lower than – your server-based statistics. The most likely causes:

  • WebTrends and other server-based software typically uses server log files (that is and GA uses tagging JavaScript code in the HTML that is reported on Google’s servers. This inherently creates differences in the way numbers are reported.
  • Search engine crawlers cannot execute JavaScript (so I hear) and therefore do not show up in GA numbers. The amount of search engine spider traffic can be huge if you have a large site. Especially if it’s crawled frequently. Check out the big spikes in the page views for one of my sites in the graphic below. Search spiders seem to be the only logical explanation there.
  • iframes and AJAX applications may result in Google under reporting.
  • Views of document files (PDF, Word, etc.) may be counted as page views in log file based software, but they will not be counted in GA page views.
  • Users with JavaScript turned off with not be counted in GA. Although those people usually account for less than 5% of all users.
  • Some web managers forget to add GA’s code to every page of the site.
  • Some claim that users will often leave a page before the page (and GA code at the bottom) has fully loaded.
  • Google doesn’t like your site (or you).

Traffic spikes from search engines on a site with ~16,000 pages.

Side note – 1&1’s (the host for most of these sites) online Web statistics are painfully slow! As much business as they get you would think they would have a better stats platform. Another example of their low cost = low quality strategy.

Follow Discussion

6 Responses to “Your Website Numbers are Lying to You”

  1. VinOne

    I found that Google Analytics just isn’t worth using for total numbers. I look to it more for trends, referral traffic, etc. When looking at it for percentage trending and conversions it’s a great tool. I’ve seen cases where on a particular day one of my adsense ads on a page actually generate more impressions than the page recorded views.

  2. Darby Tonn

    Youre not the average weblog writer, guy. You definitely have some thing powerful to add towards the web. Your style is so strong that you simply could nearly get away with becoming a bad writer, but youre even awesome at expressing what you’ve to express. Such a great weblog. Ill be back for more.

  3. Link Exchange

    Love the information here. Adding to my bookmarks. Thank you!

  4. Sunday School Teacher

    I would agree with the general trends mentioned here that I have noticed on my sites. I would still like to know which is closer to reality – Google numbers or the server logs. I see the server logs are about double the numbers in google analytics.

Trackbacks

  1. Most Popular NFL Team Websites (and Least Popular) : FatPurple  
  2. 5 Tools to Quickly Measure a Website’s Popularity : FatPurple  

Leave a Reply