Google Analytics Is Underreporting Results from Twitter

Question MarkB2B marketers are undervaluing Twitter’s contribution to their business.

That’s hard to believe at a time when many B2B marketers are glomming on to Twitter like it’s the last shiny object opportunity they will ever have (even turning to spamming), but it’s true. It is happening because web analytics packages are dramatically underreporting traffic from Twitter.

How many leads came from Twitter? How many website visits? How many purchases? When your web analytics package doesn’t recognize that traffic is coming from Twitter, Twitter doesn’t get the credit it deserves.

Twitter Traffic Is Significantly Understated

The impact is big. On my site (B2Bdigital.net), Google Analytics is underreporting Twitter traffic by 30%!

Over seven weeks, underreporting peaked at more than 40% and varied widely. The chart below compares Google Analytics and Twitter’s Site Analytics reported traffic week by week.

Chart Showing Google Analytics Underreporting Twitter Traffic

Three Reasons Twitter Results Are Underreported

Years ago, there were a number of articles saying Twitter traffic was underreported by a huge margin. But Twitter’s introduction of the t.co link shortener was supposed to address the issue.

Although t.co has helped, as Twitter becomes more important for marketers, accurate measurement is more important as well. Problems persist for a number of reasons, but these three appear to be the most prevalent in my investigation:

  1. Browsers don’t always provide referring URLs. Twitter reporting relies on having accurate and complete referring URL data.
  2. You can’t rely on campaign tracking codes in social. For paid media programs, you provide the URLs with tracking strings that your analytics solution recognizes. But in social media, most traffic will come from other people sharing your content. You don’t control the URL that is shared.
  3. Other people’s campaign codes interfere with your reporting. A number of social media tools, including Buffer, one of my favorites, automatically append their own campaign codes. In Google Analytics this overrides referring URLs completely, and it is surprisingly common.

Here is an example of #3: Two weeks ago, Mike Volpe shared one of my posts on Twitter, but visits from his Tweet aren’t reflected in my Google Analytics reporting. Here’s his Tweet:

The link expands to https://b2bdigital.net/2013/11/07/b2b-marketing-twitter-spammers/?utm_source=twitter&utm_medium=social&utm_content=2700945 and reporting for Mike’s Tweet is broken out separately from other Twitter traffic.

While this example is, thankfully, easy to spot in reporting, after implementing the solution outlined below I found traffic from Twitter reported with a wide variety of labels, some of which originally appeared in Google Analytics to be emails or even different social networks!

How To Get Back (Some Of) Your Twitter Traffic Data

If you use Google Analytics, you can’t see the referring URL if tracking codes have been added to a URL. However, with a custom filter you can capture and store the referring domain for later reporting. Below is a screenshot of the custom Google Analytics filter I am using to capture this information.

Screenshot of Capture and Parse All Referring Domains in Google Analytics

The “Extract A” value is (https?://)(([^/]+)/).

This works for all traffic, not just traffic coming from Twitter. Here are three examples of ways you can use this data, in addition to getting more complete reporting from Twitter and other social sites.

  • See actual referrers for campaign traffic. For example, by selecting my email newsletter traffic and adding the new referring URL field as a second dimension, I can see traffic that was previously reported as being from email subscribers is from when those subscribers shared the post on Twitter, LinkedIn and other sites.
  • See how your campaigns are shared on each social networks. To do this, report on your new referring URL field, add Source / Medium or Campaign as your secondary dimension and filter the report for the social network you are interested in.
  • Monitor traffic trends over time with a more consistent methodology. Although browser changes will impact referring URL reporting, removing the impact of changes in the behavior of social sharing tools like Buffer or Hootsuite will make your trend data more accurate.

Your Turn

Glad you made it this far! I’d love to hear what you think of the impact underreporting may have on marketers or alternative ways to improve the accuracy of traffic reporting from Twitter and other social networks. Share your view in the comments below or with me on Twitter (@wittlake)!

Top Photo Credit: the Italian voice via Flickr cc

Get every post delivered directly to your inbox.

Your email address will not be shared or sold. I hate spam too.