Most Meta advertisers are in their account too much.
They log in every morning, check yesterday's spend, get nervous about a CPA spike, and start fiddling. They turn ads off too early. They turn ads on too late. They restart the learning phase by accident. They make budget changes based on three days of data and then wonder why their results never stabilise.
The instinct is understandable. The account feels like it needs attention. But the truth is the opposite: most accounts underperform because they're being managed too actively, not because they're being managed too passively.
What separates the agencies that scale accounts from the agencies that just maintain them isn't intensity. It's cadence. A clear rhythm. A specific week-by-week structure that tells you what to focus on right now, what to leave alone, and when to make the next move.
This is the cadence we run inside YMP-managed accounts. Six weeks, three phases, on rotation, indefinitely.
Why Cadence Beats Intensity
Imagine you wanted to get good at chess. Two coaches offer to teach you.
Coach A says: "Just play a lot of chess. Come back to me in a year when you're better."
Coach B says: "Here's a system. Every week we'll work on one specific element. Week one, openings. Week two, mid-game tactics. Week three, endgame theory. Come back to me in a year and we'll know exactly where you stand."
You'd pick Coach B. Not because Coach A is wrong, playing a lot of chess will, eventually, make you better. But Coach B's structure means every hour of practice is anchored to a specific outcome. You always know what you're working on, what you're not working on, and what comes next.
That's what the cadence does for ad management.
Without it, you'll spend three weeks panicking about CPA, two days on creative, half a week analysing, and you'll never know whether the gains you're making are real or random. With it, you always know exactly where you are in the cycle, and so does anyone else on the team.
The cadence has three phases: test, analyse and optimise, create. Each runs for two weeks. The cycle repeats every six.
The Six-Week Cycle
Here's the structure, end to end.
- Weeks 1, 2: Test. New ads go in. You leave them alone.
- End of Week 2: Analyse and optimise. Pull insights from the testing window. Turn off the underperformers. Keep what's working.
- Weeks 3, 4: Create. Use the insights from the analysis to brief and produce the next round of creative.
- End of Week 4: Analyse and optimise again. Same process, review, learn, kill what isn't working.
- Weeks 5, 6: Test again. New ads in. Cycle restarts.
That's it. Three phases. Two-week intervals. Indefinitely.
The discipline is in resisting the urge to short-circuit any phase. Each one earns its time for a specific reason, and skipping any of them breaks the whole loop.
Let's go through them one at a time.
Phase One: The Testing Window (Weeks 1, 2)
You've added a new batch of ads to the account. The phase that follows is the hardest one, because it requires the most patience.
For two weeks, you leave it alone.
That sounds simple. It is not, in practice. Every instinct will tell you to interfere. An ad will spike on day three and you'll want to scale it. Another will look weak on day five and you'll want to kill it. Spend will skew toward one creative on day seven and you'll start second-guessing the test design.
Resist all of it.
There's a reason for the two-week window. Meta's algorithm needs time and volume to actually learn who responds to your creative. Three days isn't enough. Five isn't enough. You need a full learning-phase exit on the campaign, plus enough additional spend on each creative for the data to mean something. For most accounts, that takes about two weeks.
If you make changes inside that window, you reset the learning. You corrupt the data. And when the test is over, you'll have no clean answer to the question the test was supposed to answer: which of these creatives actually works?
The one exception. If an ad is spending real money and performing at three times your breakeven CPA, or worse, you can turn it off. It's haemorrhaging. Leaving it on for the sake of "the test" is just lighting money on fire. But that's the only exception. Not "underperforming." Not "below target." Three times breakeven, or off. Everything else stays on, and you wait.
What spend distribution will tell you (and what it won't). During the testing window, Meta will skew budget toward two or three creatives. That's normal. That's actually what you want. What it doesn't mean: that the other ads are bad. It means Meta's predicted that those two or three creatives can move the most volume at the lowest cost, so it puts the budget there. The other ads may still be perfectly profitable; they just may not be predicted to scale to the same level.
This matters because it changes how you read the results at the end of week two. The question isn't which ads spent the most. The question is which ads performed at acceptable CPA, regardless of spend. Sometimes the best-performing creative on ROAS is the one that only spent £200, because it's reaching a hot but small pocket of your audience. That ad doesn't need to scale to be valuable. Leave it on. It'll do its job when called upon.
Phase Two: Analyse And Optimise (End of Week 2)
Two weeks in. The test is done. Now you read what it told you.
This is the phase most advertisers do badly, not because the work is hard, but because they don't do it deliberately. They glance at the numbers, kill the weak ads, and move on. The analysis layer gets skipped.
That's the mistake. The optimisation is the easy half. The analysis is where the leverage is.
What "analyse" actually means. Analysis is the answer to one question, asked in detail: what worked, and why did it work? The "what" is the easy part. You can sort ads by ROAS, CPA, or volume. The "why" is the part most people never do, and it's the part that determines what you brief next.
For every winning ad, ask:
- Was it the angle? Did this ad lean into a specific customer emotion or buying trigger that the others didn't?
- Was it the format? Was video outperforming static? Was UGC outperforming founder-led?
- Was it the offer in the creative? Did the ad lead with a stronger hook, a clearer mechanism, a more credible proof point?
- Was it the trend lens? Did this ad borrow a current organic format that the others didn't?
- Was it the audience match? Did Meta find a pocket of buyers this creative happened to fit unusually well?
- Was it a competitor signal? Did this ad beat a competitor's offer that the others didn't address?
These aren't theoretical questions. The answer to each one feeds directly into the creation phase that follows, which is why the analysis has to happen before the creation, not in parallel with it. If you don't analyse properly, the next round of creative is a guess. You're back to the dartboard with the blindfold on. The whole cycle starts over with corrupted inputs.
What "optimise" actually means. Optimisation is simpler than people make it. Two rules:
Rule 1: Turn off anything that's spent too much without producing acceptable results. "Too much" depends on your account size, but for most accounts, an ad that's burned through 1, 2x your target CPA without converting is done. Cut it.
Rule 2: Leave everything else on. Including the low-spend winners. Including the slow burners. The only ads that should come off are the ones the data has clearly judged.
That's optimisation. It's not finding the "single best ad" and consolidating spend onto it. That's the fastest way to crash an account, because the single best ad doesn't have the audience depth to sustain its performance at higher budgets. You need a portfolio.
Phase Three: Create (Weeks 3, 4)
You now have a clear picture of what worked, what didn't, and why. The next two weeks are for turning that analysis into the next batch of creative.
This is where most advertisers create. But there's a critical difference between creating and creating with direction.
Creating without direction. The default mode for most ad accounts looks like this: at the end of the testing window, the operator briefs a new batch of ads based on whatever ideas came to them that week. Some are inspired by competitor research. Some by a creative they saw in another niche. Some by a hunch. The new ads go in. Two weeks later, some work and some don't, and the operator briefs the next batch the same way. The result is a creative library full of unrelated experiments. No theme. No build. No compounding insight. Every cycle starts from scratch.
Creating with direction. The cadence demands the opposite. If the analysis at end of week two told you that your testimonial ads were outperforming your product-demo ads, then the next two weeks of creative are themed around testimonials. More testimonial angles. More testimonial executions. More variations on what's clearly working. If founder-led video was beating UGC, the next batch is founder-led-heavy. If a specific angle, say, addressing a single objection in the hook, was outperforming generic benefit-led hooks, the next batch leans into objection-led structures.
This is what direction means. You're not abandoning the things that didn't work; you're not putting all your spend on one execution; you're not refusing to test new ideas. You're just weighting the next batch toward the patterns the data has already validated.
Over six months of doing this consistently, your creative library starts compounding. Every cycle builds on the last one. Every winning angle gets explored in more depth. Every losing angle gets phased out. The account doesn't just maintain, it improves, structurally, every six weeks.
How much creative per cycle? There's no single right number. It depends on account size, budget, and category. But the principle: enough new creative to maintain testing volume, without producing so much that quality drops. For most accounts running Advantage+ Shopping or a similar consolidated structure, that's somewhere between 6 and 15 new creatives per cycle. Smaller accounts at the lower end, larger ones at the upper end. The mistake to avoid is producing creative for the sake of producing creative. The cadence gives you two whole weeks to make new ads, that doesn't mean you have to fill all 80 hours of that time with creation. It means you have a window to create deliberately. Use it for thinking, briefing, and producing a smaller number of high-quality assets, not for grinding out volume.
Why The Cadence Stops You Making The Most Common Account Mistakes
The cadence isn't just a productivity tool. It's a defence mechanism against the mistakes that kill most ad accounts.
Mistake 1: Testing for too short a time. Three-day tests are pointless. The data is noise. The cadence forces a two-week minimum, every time, with no exceptions outside the 3x breakeven rule.
Mistake 2: Living in the account. Daily check-ins lead to daily fiddling. Daily fiddling kills learning phases, resets data, and confuses the algorithm. The cadence gives you specific moments to be active and explicit phases to leave the account alone.
Mistake 3: Skipping analysis. The default is to optimise (kill the bad ads) without analysing (understand why the good ones won). The cadence builds the analysis into the structure, it's a phase, not an afterthought.
Mistake 4: Creating without direction. Without the cadence, every batch of new creative is a fresh guess. With it, every batch is informed by the last one. The work compounds.
Mistake 5: Panicking on knee-jerk reactions. A CPA spike on a Tuesday doesn't mean the account is broken. The cadence builds in the patience to let signals develop before reacting, which is where the actual judgement lives.
Where You Always Are
The compounding benefit of operating in a cadence is something most advertisers don't realise until they're inside one: you always know where you are.
If someone asked you right now, "what are you doing in the account this week?", could you answer?
If you're in the cadence, the answer is always one of three things:
- "I'm in the testing window. I added new ads on Monday, they've been live four days, I'm leaving them alone for another nine days, then I'll analyse."
- "I just optimised at the end of week two. Right now I'm two days into the creative window. The data showed testimonial-led ads are winning, so the next batch is themed around that."
- "I'm at end of week four. Just optimised again. Adding the new ads on Monday, then the next test window starts."
You always know. Every action has a reason. Every reason has a phase. And the structure means anyone on the team, a media buyer, a strategist, a client, can look at the calendar and immediately know what's happening.
That clarity is worth more than any single tactic. It's what separates ad management as a profession from ad management as a panic.
A Final Note On Discipline
The hardest part of running this cadence isn't the work. It's the patience.
You will watch CPAs wobble in the testing window and want to intervene. You will see a low-spend ad that looks underwhelming and want to kill it before the cycle's done. You will be tempted, at the end of week two, to skip the analysis and just brief the next batch from instinct.
Don't.
The cadence is built to absorb noise. It's designed to find signal across two-week windows, not two-day ones. Trust it. The discipline is the strategy.
