Traffic sources that contribute to direct traffic

Posted by on Sep 12, 2012 in analytics, code, howto, stats | No Comments
Traffic sources that contribute to direct traffic

It’s worth knowing what makes up direct traffic in your analytics reports. Why? I’ve been seeing a strange set of numbers in Page traffic reports in Google Analytics recently: when comparing traffic sources for the homepage, I’ve seen much higher bounce rates for direct traffic than for organic Google referrals.

Direct traffic is usually defined as traffic coming from:

  1. people typing your address directly into their browser’s url bar
  2. bookmarks, and
  3. Google search auto-completion

When thinking of direct traffic in this way, it’s hard to understand why direct visitors to your homepage would bounce more often than those finding you via natural search. Surely people who know who you are and have taken deliberate action to visit you would be more likely to get beyond the homepage?

Possibly visitors are simply coming for your phone number or street address, finding it in your header or footer and then disappearing. Or maybe you happen to own a national daily newspaper and visitors are just scanning the headlines before leaving again. These behaviours might explain some of the direct traffic bouncing but surely not in the quantities I’ve seen.

As it turns out, Google Analytics, actually counts many more sources of traffic as “direct” than you’d think. In fact:

“…the term ‘Direct Traffic’… Google uses as a catch-all category of traffic that they simply cannot accurately measure the source of.” Wayne Lambert

Fortunately there’s a lot you can do about it. By understanding why Google can’t measure the source you should be able to take action to correct it. So, what other links can count as direct:

  • Links from PDFs and Word docs
  • Links from offline apps (e.g. Outlook, Tweetdeck, etc.)
  • Missing tracking
  • Untracked AdWords campaigns
  • Untracked 3rd party shopping cart
  • 302 directs
  • Server stripping the gclid
  • Javascript encoded links

Obviously broken or missing links and IDs should be fixed, but how about those links from PDFs, Word and other offline apps? Step forward the Google URL Builder. Simply by creating custom links via the URL Builder, you can easily share urls that will make much more sense in Analytics later.

In the following example:

Bounce rate, when you should ignore it and when you shouldn’t

…I’ve created a link to my earlier article on bounce rates that I can share via Twitter. Breaking the url down, you see at the end the three attributes GA uses to differentiate this link:

  • utm_source=twitter
  • utm_medium=social
  • utm_campaign=BounceRate

Using this bespoke link I’ll be able track visitors from Twitter much more effectively. Furthermore, these Twitter visitors will now be removed from the “catch-all” direct traffic pot, thereby making that collection of visitors much more accurate as well. Double win!

Leave a Reply