Overnight, Facebook released the statement that they’ve identified and fixed a bug that misattributed clicks on video carousel ads when viewed via a mobile web browser.
Essentially, if you were bidding on link clicks, and users expanded your video from within a carousel to watch it fullscreen on mobile, this was inaccurately reported as a link click, and you have been billed accordingly. The non-clickbait truth is that Facebook estimated this impacted 0.04% of all ads impressions (ie: not many), but as we would expect, they’ve taken it seriously and resolved the bug.
Facebook also mentioned that advertisers will receive full credit for the charges they incurred for these misattributed clicks. So all’s well that ends well???
Is looking at multiple sources of data the solution?
What this brings to light is a wider discussion about how we track, report, and make decisions on paid campaigns. Any marketer who has historically compared Facebook to Google Analytics (or any other second data source) throughout a campaign has probably attempted to quantify major discrepancies.
Internally we always look at multiple sources of data (ideally at least 2), and often rely on Facebook exclusively for the amount billed. We’ll then look to a second channel to see the more tangible result (visits/leads/add to carts/sales/etc), and then it becomes a simple equation.
If we can see the direct impact on Google Analytics (just say), and divide that by the amount billed (from Facebook), then we know exactly what our cost-per-result was. Which means we are able to report with near 100% accuracy every time.
Now, here’s where it becomes complicated.
What do you do when you’re gunning for video views? Or making a decision based on what creative is driving the most engagement? We’ve got to rely on the ad networks to provide accurate data, even if just indicative.
Generally speaking, we’ve found that you can count on Facebook data to see trends with some degree of accuracy, but if you need to optimise for an objective off Facebook (e.g. website conversions), then adding in multiple sources of tracking can only help you!
Any questions, thoughts, or ideas on how to do this better? Let me know @matthewkelly.