The Facebook learning phase is a period at the start of a new ad set (or after a significant edit to an existing one) during which Meta's ad delivery system is gathering data and optimising how it shows your ads. During this period, performance is typically less stable and less efficient than it will be once the system has collected enough data to understand who responds best to your ads and when.
How does the learning phase work?
Meta's ad delivery system uses machine learning to determine the best audience segments, placements, and times of day to show your ads in order to achieve your campaign objective. When an ad set is new, the algorithm has no historical data to work from and must experiment to find patterns. Each time your ad is shown, the system learns something about who engages with it.
The learning phase ends once the ad set has recorded approximately 50 optimisation events within a seven-day period. An optimisation event is the action your campaign is set to optimise for, such as a purchase, a lead, or a link click.
How long does the learning phase last?
The learning phase typically lasts up to seven days. If the ad set achieves 50 optimisation events within that window, it exits the learning phase and enters a more stable delivery phase. If the ad set does not reach 50 events within seven days because the budget is too low or the audience too narrow, it may remain in or return to the learning phase. Ad sets that never exit the learning phase have consistently higher costs per result.
What triggers a new learning phase?
Significant edits to an ad set reset the learning phase because the changes alter the delivery system's assumptions and require it to relearn. Actions that trigger a reset include:
- Changing the bid strategy or bid cap
- Changing the budget significantly (large increases or decreases)
- Changing the audience targeting
- Changing the ad creative (adding, removing, or editing ads within the ad set)
- Pausing the ad set for a significant period and reactivating it
How should you manage campaigns during the learning phase?
The primary recommendation is to avoid making significant changes during the learning phase, as each edit resets it and prolongs the period of unstable performance. Set your budget at a level that gives the ad set a realistic chance of reaching 50 optimisation events within seven days before launching. If performance looks poor during the learning phase, resist the urge to edit immediately: wait until the learning phase ends before assessing results or making adjustments. Consolidating ad sets rather than running many small ones also helps reach the event threshold faster.
Managing the learning phase well, avoiding unnecessary edits and consolidating conversions, is a core part of professional Facebook Ads management.