# Learning Phase

> The initial period after a campaign or ad set launches or changes, during which the ad platform's algorithm gathers conversion data before delivery stabilizes.

The **learning phase** is the initial period after a campaign or ad set launches or is significantly edited, during which the ad platform's algorithm gathers conversion data before delivery and cost stabilize. On Meta, an ad set is typically considered out of the learning phase after roughly 50 optimization events in a week; performance is usually less stable and more expensive until then.

## How AnyTrack helps exit the learning phase

The learning phase ends faster and more reliably when the platform receives a clean, steady stream of the conversions it is optimizing for. By sending server-side conversions through the [Conversion API](/glossary/conversion-api), AnyTrack restores the events ad blockers and iOS would otherwise drop — giving the algorithm enough signal to exit the learning phase instead of stalling in "learning limited" on sparse data.

## How the learning phase relates to objective and bidding

- [Campaign Objective](/glossary/campaign-objective) — defines which conversion event the learning phase must accumulate.
- [Bidding Strategy](/glossary/bidding-strategy) — does not deliver stable cost-per-result until the learning phase completes.