What if I told you that
— Jess @ FireTeam 🔥 An agency on fire... (@HireFireTeam) November 2, 2023
Meta's new "feature" which tells you how much you can raise the budget before entering the learning phase (usually by up to 70%) is just a mind game to get you to spend more because the "learning phase" doesn't exist and never has.
Would you believe me?
Tell me more…
The argument, at least from my point of view is that there is no on/off state where an ad is in the learning phase. Meta has conditioned media buyers to accept a certain level of instability for new ads and for budget changes by calling this period "learning phase". But that…
— Jess @ FireTeam 🔥 An agency on fire... (@HireFireTeam) November 2, 2023
I mostly agree with this but it’s a bit over the top (it is on X so guess that’s not a surprise).
Machine learning algorithms have a training stage. A set of training data is used to “re-wire” a model to adapt the outputs. And can be done without overwriting prior states.
This is (partially) why you should wait a week or two before deciding if your campaign is working.
At Meta scale, more data is better. So that learning label is just saying you haven’t hit Meta scale. Your performance may not be optimally optimized but let your data tell you if it’s optimal enough.