Activation & operations
Learning phase
The learning phase is the period after launch or significant edits when a platform's delivery algorithms are still calibrating who to serve and at what cost. Performance is volatile and unrepresentative during learning, and frequent edits restart it, trapping campaigns in permanent instability.
In practice
Practical implications: give new lines enough conversion volume and budget to exit learning, batch edits rather than tinkering daily, and discount the first days' numbers when judging performance.
Budgets spread across too many ad sets are a structural cause of lines that never learn - consolidation is often the fix.