Facebook Attribution: Smoke & Mirrors or the Real Deal?

Whilst it is tempting for any current blog regarding Facebook to address the ongoing scandal around the illicit harvesting, packaging and selling of unknowing users’ data, there is another Facebook practice shrouded in just as much mystery: attribution.

Many digital marketing channels have moved their attribution towards time-decay models, data-driven attribution or, at the very least, last click. However, Facebook stubbornly remains in a siloed, insular attribution model that is more in keeping with the early 2000’s than 2018. Ultimately, this is impacting advertiser’s desire to invest further in a channel that is unparalleled in terms of its targeting capabilities and creative options.

The Problem

The main problem with Facebook’s attribution is their steadfast refusal to allow third-party post-view tracking on all custom targeting options. Without third-party tracking, it’s not possible to de-dupe against your other marketing channels. As a result, Facebook will attribute a sale to an ad within Facebook regardless of whether this was the only touch point in the path to purchase or whether the user viewed a display ad, clicked a search ad and then just glanced at their Facebook during the checkout process.

This effect can be seen in the example below, where three channels would have claimed responsibility for the driven sale.

Anecdotally, I’ve heard stories of Paid Social claiming up to a ridiculous 50% of total digital revenue as a result of this siloed view, with over 80% of overall revenue already accounted for by properly attributed channels.

Consequently, we as marketers are put in a position where we need to decide between completely overvaluing the effect of post-view interactions on revenue, or solely utilising post-click attribution via attribution services such as Google Analytics.

This overvaluation of Facebook’s role in the conversion path ends up having the inverse effect, causing advertisers to undervalue Paid Social. A lack of faith in results driven through this channel means that even the sales genuinely driven via Paid Social are often lost in the crowd of misallocated conversions. Ultimately, this leads to inefficient budgeting and a serious lost opportunity to show compelling advertising to billions of active daily users.

There’s Another Way

Many marketers have decided to continue to use Facebook’s massively flawed attribution “model” (i.e. award everything to Facebook) rather than look at alternatives. This is clearly unacceptable. How can we begin to apply bid optimisations, determine budget allocation and effectively test using such an inaccurate system? We’ve lacked a solution to efficiently attribute sales involving a Facebook/Instagram touch point.

One advantage we have in our aim for clearer, more accurate Facebook attribution is that our accounts are all whitelisted to use third-party post-view on non-custom Facebook audiences. This includes all interest, behavioural or demographic targeting, but excludes anything built from CRM lists or the Facebook pixel, including look-a-likes and all dynamic activity.

By being whitelisted, we are able to appropriately attribute and de-dupe post-view sales from these campaigns, but also to view the role these campaigns play in the path to purchase. Whilst a video prospecting campaign may not lead to many de-duped post-view sales, it will play a key role in the path to purchase. Without insight into brand engagement, budgets may be unfairly cut.

Whilst Facebook are reluctant to allow full insight into post-view ad performance, they are much more open with post-click via the use of click trackers from DoubleClick, Google Analytics, and affiliate networks. However, many advertisers have yet to make use of the click trackers, potentially because they fear that the de-duped view will drastically devalue their channel or misunderstanding the effort involved in setting this up.  If your reluctance to de-dupe is born from fear of the results then, put simply, you either need to have more faith in your campaigns or improve them!

We’ve seen strong results when de-duping on a post click basis for direct response campaigns. For example, in the campaign below, despite losing 70% of Facebook-attributed post-click sales when de-duped, we were still able to drive revenue at a ROAS of over 400%.

The level of de-duping varies quite significantly across the various campaign strategies, but this gives an idea as to the overall scale of sales Facebook will mistakenly claim 100% of the credit for. This is also discounting the vast amount of additional post-view “sales” that Facebook will claim. De-duping avoids the double or even triple counting of sales seen in the path to purchase shown earlier. Instead, by utilising third-party attribution tracking, we are left with one channel claiming the purchase whilst appreciating the value of other touchpoints.

To Conclude

The ultimate result of this is that your Facebook and Instagram activity becomes a bona fide part of your marketing funnel. You can have faith in the sales that Facebook claims and use insights into the path to purchase to efficiently allocate prospecting budget.