Free tools
How much do you need to spend before Meta optimises properly?
Meta needs roughly 50 optimisation events per ad set per week to leave the learning phase. That makes your minimum budget a calculation, not a guess — it depends on what one result costs you and how many ad sets you are splitting spend across.
Inputs
What you already know.
Outputs
What it actually takes.
Fill in your numbers to see whether your budget can clear the learning phase.
How this works
The maths, in one line
An ad set leaves the learning phase after about 50 optimisation events in a rolling seven-day window. So the weekly budget one ad set needs is 50 × your cost per result, and the daily budget is that divided by seven. Multiply by the number of ad sets, because progress is counted per ad set and they do not pool.
This is why splitting a modest budget across many ad sets so often backfires. Three ad sets on $100/day behave far worse than one ad set on $100/day, not because the audiences are wrong but because none of the three ever gathers enough signal. If the numbers above look impossible, the cheapest fix is almost always fewer ad sets — not more money.
One caveat worth stating: this models the budget requirement, not the outcome. It assumes your tracking actually reports conversions back to Meta. If the pixel or Conversions API is misconfigured, Meta never sees the events, the ad set never exits learning, and no budget fixes it.
Questions
Learning phase FAQ
How many conversions does it take to exit the Meta learning phase?
About 50 optimisation events per ad set within a rolling seven-day window. An optimisation event is whatever you picked as the conversion goal — a purchase, a lead, an app install — not a click or an impression. The window rolls, so the 50 have to land inside any given seven days rather than accumulating slowly over a month.
What happens if my budget is too low to get there?
The ad set can sit in learning indefinitely, and Meta may mark it "Learning Limited" and reduce delivery. Performance stays unstable and your cost per result usually looks worse than it eventually would. Splitting a small budget across several ad sets is the most common cause — each one needs its own 50 events, so three ad sets need three times the volume.
Is it better to raise the budget or cut the number of ad sets?
Usually cut the ad sets. Consolidating spend into fewer ad sets reaches the threshold faster without spending more, and it gives Meta a bigger pool of signal to optimise against. Raising the budget works too, but it costs more and large budget changes mid-flight — over roughly 20% — reset the learning phase and start the count again.
Does the learning phase restart if I edit the campaign?
Significant edits do restart it: changing the budget substantially, changing targeting, swapping the bid strategy, or changing the optimisation event. Smaller changes such as renaming an ad set generally do not. This is why it pays to get the setup right before launch rather than fixing it live — every correction after launch costs you the learning period again.
Before you launch
Run a free check
Budget is one of eight things that decide this.
A budget that clears the learning phase still will not save a campaign with broken tracking, a mismatched objective, or an audience too small to deliver. Spendifier checks all eight before you spend, and scores the campaign out of 100.
Other free tools: Meta ads spend calculator. Or read which Meta ad metrics matter.