Media Puppies
Meta learning phase calculator
An ad set stuck in learning pays exploration prices forever. This works out whether your budget can produce enough conversions to get out, and what splitting it across ad sets really costs you.
Set from your location. The maths is the same in any currency — only the labels change.
Cost per purchase, lead or whatever you optimise to.
Count ad sets, not campaigns. Each one learns separately.
Meta’s stated threshold is about 50. Lower it only if you know your account behaves differently.
Runs entirely in your browser. Nothing you type is sent to us or stored anywhere.
Why the learning phase costs you money
When an ad set is new, or when you change it meaningfully, Meta does not yet know who responds. It spends part of your budget exploring. Delivery is volatile, cost per result is higher than it will settle at, and the numbers you see are not a fair read on the creative.
Meta exits that phase once an ad set has gathered enough recent conversions — the figure quoted is around fifty in a week. Below that, it stays in exploration indefinitely. That is what “learning limited” means in the delivery column.
The arithmetic nobody does
Take the monthly budget, divide by cost per conversion, convert to a week, then divide by the number of ad sets. That is conversions per ad set per week. Compare it to fifty. Most accounts under pressure fail this test badly, and the cause is almost never budget — it is that the budget was split six ways for control.
Six ad sets on a modest budget is not six tests. It is six things that never finish learning.
What to do when the number is short
- Consolidate first. It is free. Fewer ad sets, same money, more conversions each.
- Optimise to an earlier event. If purchases are too rare, optimise to add-to-cart or a qualified lead instead. You trade some precision for enough volume to learn from.
- Then, and only then, raise budget. And raise it in steps of roughly 20% — a larger jump restarts learning and costs you the days you were trying to save.
What resets it
Significant budget changes, new creative, audience edits, changing the optimisation event, and turning an ad set off and on again. Batch your changes to a schedule rather than adjusting daily, and give each change long enough to be judged.
If the budget looks sufficient but spend is not keeping up, check the budget pacing calculator — under-delivery and learning problems look identical in a weekly report.
Questions
Is 50 conversions a week an official number?+
It is the figure Meta has published as a guide, not a hard switch. Accounts vary, and some stabilise below it. Treat it as the planning threshold rather than a law.
Does this apply to Advantage+ campaigns?+
The same principle applies — automated campaign types still need conversion volume to learn from. They tend to handle thin volume slightly better because there are fewer separate learners.
Should I count ad sets or ads?+
Ad sets. Learning happens at ad set level. Several ads inside one ad set share its learning.
What if my cost per conversion is unknown?+
Use your best estimate from a similar campaign, or work backwards from what you can afford to pay for a customer. The output is a planning figure, not a forecast.
Why does splitting budgets feel like it works?+
Because it gives you control and clean comparisons. It also guarantees each ad set pays exploration prices. On a large budget you can afford that; on a small one you cannot.
Embed this calculator
Free to use on your own site. Copy this into your page — the attribution line is the only condition.
The paragraph below the iframe is the part that matters: an iframe's contents are served from our domain, so a link inside it is not a link from your page. The paragraph is. The small script just resizes the frame to fit your column — drop it if you prefer, the fixed height still works.
Other free tools
Want someone to run the campaigns behind these numbers?
We run Meta and Google ads for businesses where one customer is worth real money. You will get a straight answer on whether paid works for you — including “not yet”.
Get a free lead plan →