Messaging Apps, Google Analytics and Dark Social - Fresh Egg Blog

Messaging Apps, Google Analytics and Dark Social

Digital marketing apprentice

In this blog post we will show you how to quantify ‘dark social’ and how you can report on it.

We will also teach you about an advanced filter that you can use to re-write referral traffic from social platforms and improve the quality of your data in Google Analytics.

Users are now spoilt for choice when it comes to which social messaging app to use to message not only friends and family, but also brands and businesses for free. Chat and social messaging apps now have a higher retention and usage rate than any other mobile application.

Messaging apps now have more users than social networks

According to a report by the Business Insider, Messaging Apps have now surpassed social networks in number of active users.

However, when it comes to analysing the traffic from these apps to your website, the subject of ‘dark social’ becomes an ever-increasing issue.

The problem with measuring dark social

‘Dark social’ is an industry term for traffic that originated from a social platform but is labelled as direct traffic in your analytics platform such as Google Analytics. Not ideal.

Direct traffic in Google Analytics is generally understood to be users visiting your site directly by either typing your URL into the browser or by clicking on a bookmark.

But this isn’t the whole story – the direct traffic bucket contains all traffic that arrives to your site without a referrer. So, a much more suitable label for this bucket would be ‘Source unknown, could be direct’ – but that’s not very sexy is it?

Dark social traffic (aka misattributed social) has been becoming more of a problem as our clients need to know exactly where their traffic is coming from and having the majority of social messaging traffic labelled as direct is far from ideal.

We needed to know exactly which social messaging apps were hiding in that ‘direct’ bucket, in an attempt to bring some light to dark social we decided to run a few tests using the traffic to freshegg.com.

We started by taking a closer look at the landing pages of our direct traffic

We quickly realised that many of the blog posts, with very long URLs, made it unlikely for them to be typed into the browser by the user.

The only way these URLs could be hit directly was by users having them bookmarked within their browsers, which would mean they were 100% returning visitors. However they were not, they were 80% new users.

This suggested that they had found a link to this deep section of our site elsewhere with exception of the home page and maybe the careers page.

To quantify the issue we created an advanced segment

We set up an advanced segment which included only direct traffic but excluded sessions landing on the homepage or the careers page.

We compared the traffic that was now suspected to be coming from non-direct traffic sources with the overall direct traffic, to give us an idea of how big the issue really was.

Over 50% of our “direct traffic” was actually unlikely to be direct traffic

We then wanted to know exactly which traffic sources, and in particular which social messaging applications, would be hiding within that group of visits.

We created a test view in Google Analytics with an IP inclusion filter so it would only show traffic from our offices. We then opened links from within the messaging apps and observed how they would appear in GA’s Real-Time reports.

As expected, most of the apps we tested appeared with a traffic source/medium matching (direct)/(none) – but not all of them, the test had a few positive surprises in store.

Apps showing a valid source/medium footprint in Google Analytics

Facebook Messenger is one of the most popular messaging apps in the world. We were relieved that the traffic was visible to us - shown to be source: Facebook, medium: social.

Traffic from Facebook can also appear as a referral in your Google Analytics and there are a number of reasons for that which are explained in detail by Optimize SMART in their blog post focused specifically on Facebook dark social, but if you want a summary then what it really boils down to is whether or not your site is HTTPS!

Back to social messaging apps: WhatsApp, Skype Messenger and WeChat are the other big ones and we were really disappointed to see that they made no efforts to provide GA with a referral source.

So, without further ado, here are the traffic sources that appeared as ‘(direct)/(none)’ in Google Analytics, which is what we consider to be ‘dark social’.

Apps that appear as ‘direct’ traffic in your GA when they shouldn’t

  • WhatsApp
  • WeChat
  • Skype
  • QQMobile
  • Yahoo Messenger
  • SMS (text message)
  • Telegram
  • Signal
  • Blackberry Messenger
  • Snapchat
  • Line
  • Textnow
  • KakaoTalk
  • Trello
  • Slack

Only 7 apps out of the 22 tested appeared as something other than direct and none. Most of these being successful well known social networking platforms such as Facebook, Twitter and LinkedIn.

Viber is an excellent example of a clear source/medium when it comes to analysing social messaging traffic. Viber appeared in Google Analytics as source 'Viber' and medium 'chat'.

You may notice that this is the only app that appears as ‘chat’ rather than social, referral or none. Excellent!

However, a number of the aforementioned apps and platforms send traffic through as referral. This is not ideal but we can work with it by implementing an advanced filter to re-write the referral traffic from these sources to ‘social’.

Use an advanced filter to re-write referral traffic to social

Here is a step-by-step guide to add a filter that re-writes traffic from various Facebook sources, LinkedIn and Twitter. You can of course extend that list if you find other social referral traffic in your GA account.

  1. Navigate to the Filters section in your Test view (best to test these things before setting them live)
  2. Hit the red ‘Add Filter’ button and select the options ‘Create new Filter’ and ‘Advanced’ at the bottom of the list. Don’t forget to name your new filter
  3. Under Field A -> Extract A select campaign source
  4. Then paste the regex ((m|l|lm)\.|^)facebook\.com|^linkedin\.com|lnkd.in|t\.co$
  5. For Field B -> Extract B select campaign medium and type ‘referral’
  6. Under Output To -> Constructor select campaign medium again and type ‘social’
  7. The ensure ‘Field A Required’, ‘Field B Required’ and ‘Override Output Field are all selected
  8. Hit ‘Save’ and return in a couple of days to see if the filter has worked as expected
  9. Publish filter to your reporting view once you’re comfortable with how it changes your data
  10. Annotate all changes in all relevant GA data views

Dealing with ‘dark social’ effectively

We have social messaging apps that show up correctly in Google Analytics, and others that show up but are misattributed to the wrong channel. Additionally, there are those apps that don’t show up at all but are instead hidden in the big ‘Direct traffic’ bucket.

Another thing to consider is campaign tagging which can help identify traffic from your social campaigns.

We wrote a very detailed post about it which you can find here, and have also created a campaign tagging excel tool that will make tagging and managing your social campaigns child's play.

We would love to hear from you if you have found better way of dealing with dark social traffic from social messaging apps. Or if we have missed the latest social app from our list below then feel free to reach out or leave us a comment.

If you would like help with your Google Analytics set up or reporting – get in touch today.

Contact Us


Share this post


comments powered by Disqus