If Facebook is a major channel for your marketing, whether sponsored posts or normal, then you’re underestimating the visits and sales it brings.
The problem is that Facebook doesn’t play nicely with Google Analytics, so some of the traffic from Facebook mobile app comes as a DIRECT visit. That’s right – if a Facebook user clicks on your post on their native mobile app they won’t always appear as a Facebook social referral. This traffic is ‘dark Facebook’ traffic: it is from Facebook, but you just can’t see it.
Since around 40% of Facebook activity is on a mobile app, that means the Facebook traffic you see could be up to 40% less than the total.
Facebook hasn’t shown much interest in fixing the issue (Twitter fixed it, so it is possible), so you need to fix this in your own Google Analytics account.
Here are three approaches:
1. Basic: use campaign tagging
The simplest way to fix this, for your own posts or sponsored links on Facebook, is to attach UTM campaign tags to every link.
Google provides a simple URL builder to help.
The essential tags to add are “utm_source=facebook.com” and “utm_medium=referral”. This will override the ‘direct’ channel and put all clicks on that links into the Facebook referral bucket.
Beyond that, you can add useful tags like “utm_campaign=events_page” so you can see how many click through from your Facebook events specifically.
2. Moderate: use a custom segment to see traffic
What if much of your traffic is from enthusiastic brand advocates, sharing your pages or articles with their friends? You can’t expect them to all use an URL builder.
But you can make a simple assumption that most users on a mobile device are not going to type in a long URL into their browser address bar. So if the user comes from a mobile device, and isn’t visiting your homepage (or a short URL you deliberately post), then they are probably coming from a mobile app.
If your website is consumer facing, then the high probability is that that mobile app is Facebook.
So we can create a custom segment in GA for traffic which
(a) comes from a mobile device
(b) does not have a referrer or campaign (i.e. direct)
(c) does not land on the homepage
To start you need to create a segment where source contains 'facebook'. Then add the 'Direct mobile, not to homepage' segment:
Next, you can create a custom report to show sessions by hour:
You should see a strong correlation, which on the two web properties I tested on resulted in doubling the traffic I had attributed to Facebook.
3. Advanced: attribute micro spikes to Facebook
Caveat: you’ll need a large volume of traffic – in excess of 100 visits from Facebook a day – to try this at home
The final trick has been proved to work at The Guardian newspaper for Facebook traffic to news articles. Most Facebook activity is very transitory – active users click on a trending newsfeed item, but it quickly fades in interest.
So what you could do, using the Google Analytics API, is look for the ‘micro spikes’ in referrals that come from Facebook on a minute-by-minute basis, and then look at the direct mobile visits which came at the same time, and add these direct spikes to the total Facebook traffic.
I've played around with this and it's difficult to get right, due to the sampling Google applies, but I did manage to spot spikes over around 5 minutes that had a strong correlation with the underlying direct mobile traffic.
Could these approaches work for your site? I'm interested to hear.