How Google Gets Search Traffic Wrong

April 15, 2013

How much traffic does your website get from organic search?

Easy question, right? If you use Google Analytics (which is likely, as GA is installed on almost 60% of all websites), just click on Traffic Sources > Overview in the left sidebar menu, and you’ll see something like this:

Google Analytics - Traffic Sources

The problem is—this is almost certainly baloney. Hooey. Poppycock. Nonsense. BS.

For lightly trafficked sites, this chart may actually be quite close, within one percent or so. But for larger sites with more substantial visit counts, this figure can vary significantly from reality.

While researching how best to measure and categorize backlinks for WPOinc’s web presence optimization (WPO) Metrics Dashboard, we found that the way Google categorizes referring sites driving traffic is a bit, shall we say, misleading.

At first glance, determining how many different search engines drive traffic to your site (or rather, how many different search engines Google thinks are supplying visits) is an easy process: in the left sidebar menu in GA, click Traffic Sources > Sources > Search > Organic > Overview, then in the small menu in the center of the screen click Sources.

GA will list “all” of the different search engines producing visits to your site, normally somewhere between eight and 12 sources; 16 is the highest we’ve seen. The list will usually contain well-known sites like Google, Yahoo!,  Bing, AOL, Ask, Search.com and Babylon, along with perhaps Baidu (China) and Yandex (Russia).

However, the actual number of search engines driving traffic to your site is likely higher—possibly much higher. In the case of one larger client website, which gets tens of thousands of visits per month, GA reported that the site had received visits from 24 different search engines in the past year. But in analyzing the specific URLs supplying visits to identify alternative search engines in the mix, the actual figure turned out to be … 246. More than 10 times the figure GA had reported as “organic search” sources.

To check this for yourself, in GA, click Traffic Sources > Sources > All Traffic. At the lower-right of the screen, set Show Rows to its maximum value. You’ll see a long list of individual URLs; how long depends on many factors, including overall traffic volume (for this client site, the list contained almost 1,400 URLs). Near the top of the list (assuming the default sort: by volume of visits, descending) you’ll likely see the major search engines, industry news and association websites, partner sites, popular social networks, and large online directories (such as YellowPages.com, Hoovers, and ZoomInfo).

But scroll a bit further down the list and you may see URLs you don’t recognize. Among these are blogs, lesser-known directories, and … search engines. Here are a just a small sample of the search engines we found miscategorized by GA:

blackle.com / referral
claro-search.com / referral
crawler.com / referral
endlessmatches.com / referral
fastestwebsearch.com / referral
goodsearch.com / referral
hotbot.com / referral
i-mysearch.com / referral
metacrawler.com / referral
pipl.com / referral
qiye.us / referral
safe-find.com / referral

Granted, these sites individually don’t (normally) provide large volumes of traffic, and even collectively in our experience usually skew results by no more than 2-3%. The errors, however, can be much worse in some cases: in one recent situation, GA reported that a site’s organic search traffic had increased by less than 1% on a quarter-to-quarter basis; but an in-depth look at “referring” search engines revealed that actual organic search traffic growth was nearly 20% for that period.

WPOinc is well aware that data quality is a problem in web measurement, though one might expect an organization with Google’s resources and brainpower to get something this basic right. This example is just a subset of a larger web data quality problem, that of accurately categorizing backlinks. That is, of all the sites linking to and/or sending visits to your site, what share are blogs vs. news sites vs. search engines vs … all manner of other sites? How does that profile compare to your top competitors?

SEO tools expert Ann Smarty has reviewed backlink categorization tools, but none offer much flexibility in terms of categories and all seem to suffer from some degree of data quality issues, which is why WPOinc needed to unravel the backlink blackbox. Over time, we expect third-party tools will improve in this area, providing marketers with better data to support web presence and search engine optimization efforts—we just couldn’t wait for them. When it comes to web presence optimization (WPO), you must be able to accurately measure and categorize backlinks and be able to benchmark your backlink strategy against your competitors to get a competitive edge.

So, just a heads-up: figure that your site is probably getting more (possibly a lot more) organic search traffic than Google Analytics is telling you.

 

Post to Twitter

5 Responses


  1. Jason 

    I’ve also seen another issue with the way Google tracks SEO traffic in that it will incorrectly attribute paid search traffic to organic, even with 100% correct UTM tracking sources in place.

    I’ve been able to replicate this in multiple profiles and it is very, very aggravating to get this type of spill over. I absolutely love that Google Analytics is free and they support it pretty well, but that’s one glaring error that’s been around some time.


  2. Tom 

    Hi Jason – I’ve seen that happen too, though in my experience the misattributed numbers tend to be very small…maybe a handful of visits out of several hundred. Have you seen larger magnitude errors?

  3. Hi Jason,

    You can actually configure Google Analytics to recognize additional search engines. It’s a feature that’s been around for a long time, but often people do not use it. You can read more about it here:

    https://developers.google.com/analytics/devguides/collection/gajs/gaTrackingTraffic?hl=en#searchEngine

    I should mention that the implementation is different for the new Universal Analytics.

    Full disclosure, I work on the Google Analytics team.

    Thanks for your post,

    Justin


  4. Tom 

    Hi Justin – thanks for the clarification. It’s good that GA enables users to add alternative search engines, but in checking out that link, I can see why this isn’t widely used. It just seems like a capability that should be part of GA. But again, I appreciate your comment and the additional information. GA is the best at what it does for the price, it just isn’t perfect.

Leave a Reply