Our Weekly Meta Ads Optimisation Routine

Most Meta accounts don’t underperform because of a bad strategy. They underperform because nobody runs the same disciplined loop every week. Optimisation isn’t a burst of activity when results dip; it’s a routine you repeat whether the account is up or down.

Below is the Meta ads optimisation routine we run each week, restructured as a copyable SOP. It tells you not just what to do, but how often to do it, how much data you need before you act, and the thresholds that trigger a decision. Treat the numbers as directional benchmarks to calibrate against your own account and margins, not universal laws.

The account structure this routine assumes

The routine only makes sense if your account is built to support it. We split the account into two jobs:

• A testing campaign, usually run ABO (ad set budget optimisation), where each new concept gets its own ad set and a fair, controlled budget. This is where new creative earns its place.
• A scaling campaign, run with Advantage+ campaign budget (Meta’s current name for what used to be called CBO), where proven winners get consolidated and the algorithm distributes spend to the best performers.

If your structure doesn’t separate testing from scaling, sort that first. We cover the reasoning in the only Meta ad account structure you need to scale profitably and in the smarter way to test and scale Meta ads: ABO first, CBO later.

The weekly Meta ads optimisation routine

Step	How often	What to check	Threshold / trigger	Action to take
1. Cut underperforming ads	Weekly (glance daily)	Cost per purchase vs target CPA, after a minimum spend	Ad or ad set is ~20% above target CPA after spending at least 2–3x your target CPA (enough data to judge), or has spent ~1x target CPA with zero conversions	Pause it and reallocate the budget to what’s working. Don’t cut on a single quiet day.
2. Feed the testing campaign	Weekly	Your creative pipeline	Every week, regardless of how current ads perform	Launch 2–3 fresh concepts into the ABO testing campaign to stay ahead of fatigue and the algorithm.
3. Review the KPIs that matter	Weekly	Cost per purchase, ROAS, conversion rate, AOV (not just CTR/CPC)	ROAS below target or CPA above target across meaningful spend	Decide per ad set: pause, iterate the creative, or scale.
4. Check retargeting freshness	Weekly / monthly	Frequency, audience size, offer age	Frequency creeping up (roughly 2–3+ over 7 days is a directional warning) or an offer that’s been live for weeks	Rotate the creative and refresh the offer before performance decays.
5. Double down on winners	Weekly	Top-performing angles, offers and hooks	An ad comfortably beating target CPA/ROAS with meaningful spend behind it	Prioritise it: increase budget or move it into the scaling campaign while it’s hot.
6. Build new ads from proven winners	Weekly / ongoing	Winning hooks, formats and angles	A validated winner exists	Produce iterations that change one variable at a time (hook, format, angle or offer) so you learn what’s driving the result.
7. Make budget changes safely	When scaling	Ad set status and recent conversion volume	You want to increase budget on a winner	Raise by no more than ~20% at a time and avoid mid-week resets, so you don’t push the ad set back into the learning phase.

Cadence and thresholds, explained

Give every ad enough data before you judge it. The single most damaging habit in Meta optimisation is acting on noise. “Cut at 20% above CPA” is only safe once an ad has spent enough to be trustworthy, which is why step 1 pairs the CPA rule with a minimum-spend gate. Below that, you’re reacting to variance, not performance.

Respect the learning phase. An ad set needs roughly 50 optimisation events (the specific event you’re optimising for, e.g. purchases) within a rolling 7-day window to exit the learning phase, according to Meta’s own guidance. While it’s learning, delivery and CPAs are unstable, so a bad day means very little. Large edits, big budget jumps and frequent restarts reset that clock and keep you paying learning-phase prices. Scale in steps, not leaps.

Scale winners deliberately. When an ad is beating target with real spend behind it, increasing budget by around 20% every few days is usually gentle enough to avoid a reset while still compounding. For the fuller playbook, see here’s how we scale Meta ads the right way.

Iterate creative with intent. Steps 5 and 6 are where accounts either plateau or keep climbing. Don’t “tweak format, copy and visuals” all at once, that teaches you nothing. Take a proven winner and change one lever per test: the hook, the format (static vs video vs carousel), the angle, or the offer. Structured iteration is also your best defence against fatigue, which we go deep on in a simple system to beat Meta ad fatigue.

The point of a routine is that it removes emotion. Same checks, same thresholds, same day each week. Do that consistently and the account improves on autopilot instead of lurching between panic cuts and impulsive scaling.

FAQ

How often should I optimise a Meta ads account?

Run the full routine once a week, with a quick daily glance for anything genuinely broken (delivery stopped, spend spiking with no conversions). Optimising more often than weekly usually means reacting to noise and resetting the learning phase.

How much should an ad spend before I pause it?

As a directional rule, give an ad at least 2–3x your target CPA in spend before judging it on cost per purchase, or pause it once it’s spent roughly 1x your target CPA with no conversions at all. Calibrate this to your margins and average order value rather than treating it as fixed.

What does the 20% above CPA rule actually mean?

If a fairly-tested ad is delivering a cost per purchase around 20% higher than your target, it’s dragging efficiency down and the spend is better used elsewhere. The rule only applies after the minimum-spend gate above, otherwise you’ll cut ads that were simply having a quiet day.

Will increasing my budget reset the learning phase?

Large budget increases can push an ad set back into learning, which raises costs while it re-stabilises. Raising budget by around 20% at a time, spaced out, usually avoids a hard reset. Ad sets need roughly 50 optimisation events per 7 days to stay out of learning.

What’s the difference between the testing and scaling campaigns?

The testing campaign (typically ABO) gives each new concept a controlled budget so you can judge it fairly. The scaling campaign (using Advantage+ campaign budget) consolidates proven winners and lets Meta distribute spend to the strongest performers. Keeping the two separate stops testing from destabilising your reliable spend.

Post inspired by: Filipe Garcia