Reach & delivery
What is Learning phase?
The learning phase is the initial period after you create or significantly edit a Meta ad set, during which the algorithm explores who to show your ads to and delivery is less stable. On Meta an ad set typically exits the learning phase once it gathers roughly 50 optimisation events (such as purchases) within about a week — until then, performance and cost per result swing more than they will later.
Example
You launch an ad set optimising for purchases at $20 a day. For the first few days CPA jumps between $9 and $40 as Meta tests audiences and placements — this is the learning phase. Once it accumulates around 50 purchases in a week, delivery settles and CPA stabilises. Edit the budget heavily on day 5 and the ad set re-enters learning, restarting that instability.
Why it matters
Judging or 'fixing' an ad set while it's still learning is one of the most common Meta mistakes — the numbers are noisy by design, and every significant edit resets the phase, so constant tweaking traps an ad set in permanent instability and inflated costs. Ad sets stuck in 'Learning limited' (unable to reach ~50 events a week) are a warning that the audience is too small, the budget too low, or the optimisation event too rare.
How to use and improve it
Give a new ad set time and enough budget to reach ~50 optimisation events before you judge it, and resist editing budget, bid strategy, targeting or creative during learning since each significant change restarts the phase. Consolidate ad sets rather than splitting budget across many tiny ones — fragmented spend starves each of the events it needs to exit learning. If an ad set is 'Learning limited', widen the audience, raise the budget, or optimise for a more frequent event higher in the funnel.
Frequently asked questions
How long does the Facebook learning phase last?
It lasts until the ad set gathers roughly 50 optimisation events within about a week — often a few days to a week for well-fed ad sets. There's no fixed timer; it's driven by event volume, so low-budget or narrow ad sets can take much longer or never exit.
What resets the learning phase?
Any significant edit — changing the budget substantially, the bid strategy, the optimisation event, the targeting, or the creative — restarts learning. Minor changes may not, but Meta flags an ad set as re-entering learning when an edit is large enough to disrupt delivery.
What does 'Learning limited' mean?
It means the ad set can't gather the ~50 weekly events it needs to exit the learning phase and stabilise, usually because the audience is too small, the budget too low, or the optimisation event too rare. Widening the audience, raising budget, or consolidating ad sets is the fix.
Should I make changes during the learning phase?
Avoid significant edits while an ad set is learning, because each one resets the phase and prolongs unstable, more expensive delivery. Let it stabilise first, then adjust deliberately once you have enough data to trust the numbers.
Related
How does your account score?
A free Meta ads audit gives you a 0–100 score, shows wasted budget, and lists the exact next steps.
Run a free audit