Automated bidding systems need data before they can optimize. On Meta, an ad set generally exits learning after about 50 optimization events within seven days of its last significant edit; Google's Smart Bidding strategies show a 'Learning' status for roughly one to two weeks, or longer for low-volume campaigns. During this window, costs and results can swing noticeably.
Significant edits — large budget changes, new bid targets, new creative, or targeting changes — can restart learning. That's why disciplined advertisers batch changes, avoid daily tinkering, and structure accounts so each campaign or ad set gets enough conversion volume to learn quickly.
Example
A new Meta ad set optimizing for leads generates 60 leads in its first five days, exits the learning phase, and its cost per lead stabilizes.
How Sloss Digital uses it
We structure accounts to consolidate conversion volume, make changes in deliberate batches, and give new campaigns room to learn — rather than reacting to the noise of the first few days.