It’s a common problem, we open our Google Analytics dashboard and click on the acquisitions tab and look at the reported numbers for facebook and it’s a lot lower than what facebook reports. My initial response was to distrust numbers from facebook since they have a history of manipulating their users but it seems like they’ve recently been bending towards the requests of advertisers to be more transparent.

I’ve been reading up on some explanations marketers have been providing on these discrepancies, and assuming that most people want a set-it-and-forget-it approach to their analytics, those explanations may apply. But for a lot of us that don’t mind doing a bit of configuration, those discrepancy explanations may not be very accurate.

Cross Device Conversions

A common argument that is made in favor of facebook is it’s ability to track conversions across devices since it uses user sessions to track conversions while google uses cookies. Since different cookies are set for, say a browser on your laptop and another one for your phone, a conversion will only attribute the source within that device.

That may be the case in a default setup of analytics, but if you added tracking for those cookies, you can merge those cookies when a user logs into your service. As they use multiple devices to log into your service, those cookies are all identified as the same user and all of your analytics data associated with those disparate cookies can now be joined. (Session Unification aims to solve the problem of multiple users using the same device when you add a user id dimension which is nice but also may limit tracking the full journey if they’re not. You can come up with your own model to figure out how best to merge cookies with users.)

An alternative, and less definitive way of cross device tracking is by using additional dimensions. It would be easy if we could collect the ip addresses of our visitors and use that dimension to merge unique cookies across devices, but due to privacy laws (mostly in the EU), and usage restrictions set by google, that’s either frowned upon, illegal, or a breach of terms of service. However, you still have data like the lat/lon (accurate to within about a km), ISP, device details (brand, category, etc.) and if you were ambitious enough, you could use statistical models to get pretty close to associating devices to a user for cross device tracking.

Conclusion

So is it worth it to add all these configurations with Google Analytics to get what facebook asserts delivers out of the box? You bet it is. How else will you know whether facebook is accurately reporting conversion values? Admittedly this doesn’t work for impressions, but you can get pretty close with clicks (Frankly, I lean towards turning off impression conversions). It’s always better to have a secondary source of data (marketing, sales, or otherwise) from the person or business you’re buying from. It’s like believing sales reports from a business you intend to buy without verifying the numbers provided by the owner. If you have a means to get close to verifying those numbers, you would be much more comfortable making a purchase than just taking their word for it.