The problems with forced campaign budget optimisation
In September 2019, Facebook has decided that they’re going to force campaign budget optimisation (CBO) by default. This is a huge change for advertisers.
From a machine learning or perhaps reinforcement learning perspective, CBO makes sense: given x amount of data inputs, alter parameters to best optimise for a desired result (e.g. purchases). The main parameter in question is ad set budget: Facebook will distribute money to the “winning” ad set. The idea is that by doing this early, Facebook will get you better results, having “optimised” as best it can.
The obvious major flaw in this is that unfortunately, this means forfeiting all control to Facebook’s algorithms for the testing phase of your ads i.e. that phase where you’re putting money into ads to see which ads perform best. We typically want money to be spent on certain ad sets, however, without Facebook deciding this prematurely. So far, advertisers have typically done this manually, altering individual ad set budgets as best as possible. But with CBO, there’s a significant restriction on the ability to test ad sets: you can’t test an ad if there’s hardly any traffic sent towards it!
So from the prospecting side of things (i.e. ads to cold audiences), our testing becomes limited.
This also can become a rather unusual problem for retargetingcampaigns. A typical Frequency to aim for would be about 1 per day (for example, if retargeting an audience fitting under ‘ViewContent Last 7 Days’, then a frequency of 7 is the ideal result 1). But if you’re trying to get that ideal frequency for individual ad sets, how could you possibly titrate your ad set budget accordingly? You can’t.
Luckily, Jon Loomer mentioned that there is a great workaround. You can read his full post at his website (he and Andrew Foxwell are superb Facebook marketers), but I’ll mention briefly what you can do below.
When CBO eventually rolls out, at the time of writing there’s no intention for a user to be able to turn it off. However, one can try to set an ad set budget minimum and maximum, which is more or less the same thing.
When you edit the Ad Set within any given campaign, you can check out the Budget & Schedule section.
Obviously one caveat is that Facebook can’t always guarantee the minimum amount will actually be spent, but hey – at least trying to is something.
It’s hard to tell exactly how much this feature will benefit advertisers. It seems to benefit advertisers who don’t need nuance, and be bad for the advertisers who want it.
I’m all for attempts at multi-armed bandit experiments which is exactly what CBO is doing. Google got there first, but in essence, multi-armed bandit experiments is a fancy way of saying you throw money at the A/B tests that are winning, whilst the tests are being run.
But at the same time, direct control of ad set budgets means much finer control over specific metrics like Frequency, which CBO will unfortunately just make a little harder to do.
- Won’t name the exact source for this number here, but basically this was a tip from someone I knew who’s done Facebook marketing for the past 10 years and achieved 6x ROAS on average, so I’m inclined to follow her advice. ↩
Stay up-to-date with the latest information.
Scaling Facebook Ads from $0 to $10,000+/day spend – Dimitri Nikolakakis Facebook Ads Strategy Notes
Dimitri Nikolakakis - an Australian internet marketer who consults multiple 7-figure eCommerce brands for his own...