The scenario
The brand sells humidifiers, cold and flu kits and a vitamin range in Kuwait, direct to consumer, on a Shopify store with KNET, Tabby and cash on delivery at checkout. Four years in, a lean in-house team, and a demand curve that is almost entirely seasonal: a strong winter, a smaller bump after Ramadan, and quiet months in between that the team had learned to accept rather than plan around.
The ad account reflected that same lack of planning. Dozens of ad sets had accumulated across two winters of campaigns that were never cleaned up, most of them near-duplicates competing against each other in the same auction, and every attempt to raise the budget before a previous season had pushed cost per purchase up within days rather than growing the account.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own stock plus a backup supplier for peak-season stockouts, next-day delivery across Kuwait
- Team
- One in-house media buyer, one operations lead
An account structured to fight itself the one month it mattered most
The account had grown the way most seasonal accounts do: a new campaign added ahead of each winter, an old one never quite switched off, and a handful of retargeting ad sets rebuilt from scratch each time rather than reused. By the time we opened it, more than two dozen ad sets were live across Meta and Google, many targeting nearly identical audiences with nearly identical creative.
That structure is invisible in a quiet month, because the spend is small enough that the overlap barely shows. It becomes expensive the moment volume rises, because near-duplicate ad sets start bidding against each other in the same auction for the same person, and every previous winter the brand had responded to rising cost per purchase by adding yet another ad set rather than removing the ones already competing.
Creative supply told the same story from a different angle: only two proven creatives were actually carrying the account, everything else was residue from campaigns nobody had reviewed in a year. The brand did not have a demand problem going into winter — it had a structure that punished it for trying to meet the demand it already had.
What we did — the creative that carried the spend
The losers are here on purpose. A test with only winners was never a test.
Consolidating into one prospecting campaign per platform
Winner"One campaign, one budget, one auction — instead of fourteen fighting each other."
- Format:
- Account restructure: prospecting, retargeting and retention split into one campaign each per platform
This was the single change that mattered most before any budget moved. Collapsing more than two dozen overlapping ad sets into one prospecting campaign per platform let the budget learn as a unit instead of splitting its signal fourteen ways, and cost per purchase became readable for the first time in a year — before it even had more money behind it.
Budget raised in twenty-percent steps, held for a week each
Winner"Twenty percent up, hold a week, then decide."
- Format:
- Step-change budget schedule with a defined cost ceiling and a rollback rule
In every previous winter the brand had roughly doubled its budget in one move and judged the result within two days, which is not enough time for an account to relearn at the new spend level. Moving in twenty-percent steps with a week to settle each time was the difference between scaling and simply spending faster into the same ceiling.
Confirming six proven creatives before raising anything
Winner"Six proven ads live before the budget moves, not two."
- Format:
- Pre-scaling creative audit and a production sprint to fill the gap
Two creatives were carrying almost the entire account, and adding spend on top of only two is exactly how frequency climbs and cost per purchase drifts up regardless of structure. A short production sprint to reach six proven creatives before the first budget step gave the extra spend somewhere useful to go instead of piling onto the same two ads.
Adding Google Search as a second platform before Meta was stable
Lost"Two platforms, same week, same push."
- Format:
- Google Search campaign launched in parallel with the first Meta budget step
This was tried once and rolled back within days. Splitting attention across two platforms before the first one had settled at its new spend level made it impossible to tell which platform any change in cost per purchase belonged to, and it slowed the Meta consolidation from delivering the clean read it was designed to give.
What we did — the optimizations, in order
Put measurement first, before touching structure or budget
Server-side conversions confirmed on Meta and Google, one link-naming convention applied across every campaign, and blended marketing efficiency set as the number the whole engagement would be judged on.
Why: Scaling on a platform-reported return that double-counts or under-counts purchases just scales the error along with the spend. Fixing measurement before anything else meant every later decision — the budget steps, the platform add — was made against a number the brand could trust.
Consolidate the account into one job per campaign
More than two dozen overlapping ad sets collapsed into one prospecting campaign, one retargeting campaign and one seasonal retention campaign per platform.
Why: The fourteen near-identical ad sets competing in the same auction were the actual reason previous scaling attempts had failed, not the size of the budget. One job per campaign is what let the account's algorithm learn on a single, coherent signal instead of splitting it.
Confirm creative supply before the first budget step
An audit of every live creative against the target cost, followed by a short sprint to bring the count of proven creatives from two to six before spend increased.
Why: Adding budget to two winners is how frequency climbs and cost per purchase drifts up while everyone blames the platform. Six proven creatives gave the algorithm somewhere to spread the new spend instead of forcing it back onto the same two ads.
Move the budget in steps, not a season-opening jump
Twenty percent per step, held five to seven days, with a defined cost-per-purchase ceiling and an automatic rollback after two consecutive days above it.
Why: The account had a real ceiling, and the only way to find it without overshooting was in small, measured steps rather than one large jump timed to the first cold week. Doubling a budget and judging it two days later is not scaling, it is spending faster into the same wall.
Add Google Search only once Meta had settled
Google Search held back until the Meta prospecting campaign had absorbed three full budget steps and its cost per purchase had stayed under the ceiling for a full week.
Why: The one earlier attempt to add a second platform mid-scale had made the read unreadable, splitting attention and budget across two platforms neither of which was stable yet. Sequencing rather than parallelising the platforms is what kept every result attributable to a single decision.
Pre-load the winter budget instead of reacting to it
The season's step schedule and cost ceiling were set and shared with the brand three weeks before the winter demand curve historically began to climb, rather than decided the week costs started rising.
Why: Higher costs per thousand during the season peak are predictable in this category, and accepting them in advance with a planned ceiling is a completely different decision from discovering them mid-campaign and reacting under pressure. Planning ahead is what let the account keep its head during the one month volume mattered most.
Read the ceiling by structure and creative, not by spend alone
Every rollback was logged with which of the three levers — structure, creative supply or budget step size — most likely caused it, rather than recorded as a single unexplained cost spike.
Why: Cost per purchase rising as spend increases usually traces back to one of three things, and diagnosing which one it was each time is what let the brand keep pushing the ceiling higher across the season instead of stopping at the first sign of trouble.
What changed
The headline number is in the table above, and the shape of it is the point: spend went up going into winter and blended marketing efficiency held rather than sliding the way it had every previous season. That happened because the account stopped fighting itself before the money moved, not because any single ad performed unusually well.
Frequency on the prospecting campaign stayed flat for longer than in any previous winter, which is the clearest sign the consolidation and the creative sprint were doing their job together: more proven creatives meant the same audience was not seeing the same two ads on repeat as spend climbed.
The one rolled-back test — adding Google Search mid-scale — turned out to be as useful as the wins. It confirmed that sequencing platforms rather than running them in parallel was the right call for an account this size, and it is now written into how every future season gets planned.
What we would do next
First, add Google Search properly once the winter peak has passed and Meta has room to breathe, rather than during the one month it needs full attention, since search demand for named symptoms and products barely dips between seasons.
Second, start the post-Ramadan check-up push on the same step-and-ceiling structure a full month early, because this brand's second seasonal window is smaller but shaped exactly like the first one, and there is no reason to relearn the same lesson twice a year.