Facebook Fix Bug That Was Undercounting Referrer Data

Last week Facebook released a rather ‘under-the-radar’ announcement that they had fixed a bug that “caused many third-party analytics tools to undercount the percentage of their organic traffic from Facebook”. Tagged on to the end of a much larger announcement around a range of new tools and features they released for publishers, the announcement was likely lost to many.

See below for the full release:

We also recently fixed a bug that caused many third-party analytics tools to undercount the percentage of their organic traffic from Facebook. In the past, referrer data was missing from some of our outbound mobile clicks. This issue has been resolved for iOS, with a fix for Android coming soon. Publishers may see an increase in iOS mobile traffic from Facebook and a decrease in undefined sources or direct traffic.

As you can see, the announcement is very vague and there isn’t much of an indication as to how much of an impact or how much the bug was distorting Facebook referral data on website analytics, such as Google Analytics.

From experience, we have often seen fairly large discrepancies between what Facebook are reporting and what Google Analytics is reporting. To a certain degree that is to be expected as Facebook deals in ‘Website Clicks’ - The number of clicks on links appearing on your ad or Page that direct people to your sites off Facebook as a result of your ad, whereas Google Analytics (the third party analytics tool we use on our site) deals in ‘Sessions’ - The period of time a user is active on your site or app.

The key point is that Facebook will count a website click when someone clicks on your link, whereas Analytics will only count it when the user is confirmed to have landed successfully on your site – in technical terms, when the Google Analytics javascript is triggered on the page. From a mobile perspective this can cause discrepancies because of poor signal and slow loading times. Often users might click on your link and then lose patience before the site loads (and before the javascript is triggered) and thus Facebook counts it where Google Analytics does not.

However, clearly that was not the only explanation for why these discrepancies occur. It seems a little odd that this bug fix was just tagged on to the end of a seemingly unrelated post, however let’s just hope that the impact of the bug was so minor that it didn’t warrant a post for itself.