Broad targeting on Facebook took me to $500k / month in spend. Now what?

One common refrain from performance marketers in the consumer space is that Facebook’s targeting and audience development algorithms have become so sophisticated that marketing campaigns need really only target very broadly in order to achieve efficiency: with sufficient ad creative and enough scope, Facebook can quickly determine which people are most apt to take some action and thus all targeting should be left to Facebook.

The following Twitter thread articulates this point very explicitly:

The above thread is required reading for any performance marketer, and, in my opinion, the author is absolutely correct: at the scale of $100-500k per month, defining audiences very broadly and allowing Facebook to optimize its own targeting settings via pixel / in-app event feedback is the most efficient way to deploy advertising budget.

But in my experience, very broad targeting doesn’t scale much further than that level of spend on Facebook. In The power and peril of Facebook’s advertising platform, I wrote:

Facebook’s precision targeting capabilities amplify the positive feedback of the Honeymoon Phase: they allow the most relevant users to be targeted so hyper-surgically — avoiding anyone whose interest in the product can’t be measured to be absolute — that it can appear as if the product is destined for ubiquity. Some advertisers mistake the performance they see during this period as a commercial fait accompli; a testament to how the product will perform on all ad channels for all time. And in some cases, they may be right: if they’re targeting broadly on Facebook and seeing impressive results, they might be able to project performance out across the universe of digital ad channels for at least some time horizon.

Since that post was published in 2017, Facebook’s targeting optimization capabilities have become even more powerful: now, to target broadly is to target precisely, because Facebook can very quickly ascertain the profiles of the users most likely to complete some action (in most cases: monetize).

The problem with this is that targeting broadly obfuscates the mechanics of what Facebook is doing in the background: targeting precisely. So early traction with broad targeting can fool an advertiser into thinking that they are actually advertising to a much broader audience of people than is actually interacting with its ads, which can therefore lead them to believe that their total addressable market is larger than it is.

This often becomes apparent at the level of $500k per month (or thereabouts) in spend on Facebook: delivery stalls and conversion costs rise because Facebook’s optimization algorithm has segmented the very broad audience definition to conversion events that are ultimately superficial for a high-scale business: first purchase or 1- or 7-day ROAS. It is at this point that the advertiser needs to actually take the reins and direct Facebook (and the other channels it works with) with respect to targeting in order to scale spend for particular audiences via audience-specific ROAS modeling. This is accomplished through:

  1. Lookalike Audience Development: As stated in the above Twitter thread and across this site, the AEO and VO campaign strategies on Facebook are essentially just dynamic lookalike products. But Facebook can only optimize to the events to which it is exposed, which in most cases are monetization events. An advertiser has the complete set of behavioral events for every single user: it can build correlation models between monetization and any event, and on that basis it can build audience targets;
  2. Creative / Audience Pairing: Feeding Facebook’s targeting algorithm a large volume of ad creative allows it to optimize creative delivery against some conversion metric — but, again, Facebook only knows so much about a user (namely: the degree to which they monetize). An advertiser has a much richer set of data to craft into audience targets, including ROAS and source creative. In building its own audience definitions via custom audiences / lookalikes, the advertiser can segment those audiences across creative themes on the basis of ROAS in a way that Facebook can’t. By pairing creatives to audience definitions, an advertiser can potentially increase spend by optimizing creative at the level of the audience;
  3. In-app Product Optimization: the “future Growth team” works across an integrated product / marketing / analytics workflow that builds a virtuous cycle between in-app events and the audience targets that deploy them. This “wag the dog” approach to product development segments audiences within the product and exposes different versions of the product on the basis of marketing targeting (eg. Target Audience A sees Version A of the product, which has been optimized for their behavior to increase the likelihood of conversion). In order to do this properly, the advertiser needs to be able to target an audience explicitly — without that targeting capability, it can’t actually increase spend against a specific audience.

These deep funnel integrations between product and marketing require explicit audience definitions: spend is scaled for individual audiences on the basis of their specific monetization profiles. This can’t be done by merely targeting broadly and allowing Facebook to do the heavy lifting with audience development, and it’s why broad targeting has a natural upper limit with respect to budget scale. And, of course, segmented messaging and re-targeting aren’t possible with broad audience definitions, and the popularity of re-targeting with mobile advertisers has grown significantly over the past few years as device sales have flatlined and audience saturation has become a poignant pain point.

It is exactly for this reason that the DTC model has peaked, with some once-ascendant brands slashing spend or ending operations. Broad targeting can kick-start a business but it can’t sustain its growth, and DTC companies were especially susceptible to the illusion of a permanent up-and-to-the-right trajectory via Facebook ad spend.

Photo by Fabian Blank on Unsplash