The short answer: Facebook ads take 60–90 days to reach stable, scalable performance. Not because anyone is slow - because of how Meta's delivery system works. Anyone promising meaningful results in two weeks is either lucky, lying, or measuring something that doesn't matter.
When a campaign launches or changes significantly, Meta's delivery system enters what it calls the learning phase: a period where it's actively exploring who responds to your ads - testing your creative against different slices of your audience to figure out who converts. During this period, performance is volatile by design. Costs swing. Results look random. That's not failure; that's the system doing its job.
Meta's own guidance is that ad sets need roughly 50 optimization events within a week to exit learning. For a lead-gen business spending modestly, hitting that threshold takes time - and every big edit you make resets part of the clock. This is why panicked week-two changes are the most expensive mistake in Meta advertising: they keep the account permanently in its most volatile state.
Exiting the learning phase is only step one. What actually produces durable results is a cycle that can't be compressed:
Each stage feeds the next. You cannot skip to week nine's clarity without weeks one through eight's data - no matter who runs the account.
Early weeks aren't resultless - they're just measured differently. By week two, you should be able to point to: verified tracking (every lead traceable to its ad), a testing structure actually running, first kill decisions already made with reasons, and a cost-per-lead trendline forming - even if the number itself is still high. Process visibility is the week-2 milestone. Results are the week-9 milestone.
The "results in two weeks" pitch works because by the time week six arrives and nothing has stabilized, you've already burned the budget and lost patience. That's the actual function of the promise - it sells hope and buys time. So write the honest 60–90 day window down before you judge a single number: if the timeline is real, you don't need miracles by week two to know it's working. The free tools help you measure the right thing while the data matures.
Want the honest, week-by-week version? The free playbook lays out the whole 60–90 day window - what to build, test, and measure - so you can run it yourself. Need fresh creative variants to feed those tests? A tool like Vidfly spins them up fast.