The scenario
The brand sells athletic sneakers and streetwear pieces aimed at a younger Kuwait audience, on a Shopify store with Tabby and cash on delivery at checkout, shipped from its own small warehouse. It spent its first year scaling almost entirely on Meta before bringing in a freelance media buyer to add Snapchat and TikTok.
The Meta ads manager showed a healthy return on ad spend the whole time, and the founder kept raising budget on the strength of that one number. What it never showed was the cost of the pairs coming back, which by the time we looked was quietly eating close to a third of the margin the dashboard said the account was making.
- Monthly revenue band
- pending client sign-off
- Average order value
- pending client sign-off
- Fulfilment
- Own small warehouse, next-day courier, cash on delivery nationwide
- Team
- Founder plus a freelance media buyer
The dashboard said healthy. The bank account disagreed.
The account was being scaled on a return-on-ad-spend number that a footwear business specifically cannot trust at face value: it counts every sale the moment it happens and never subtracts the pair that comes back three weeks later. Frequency on the prospecting audience had also climbed past the point where a Kuwait-sized cold audience keeps responding, because every budget increase had gone onto the same two ads.
When we rebuilt the numbers net of confirmed returns and refused cash-on-delivery orders, the picture changed: the account was still growing revenue, but growing it at a return that was noticeably thinner than the dashboard implied, and the gap had been widening every month the budget went up. Scaling had not created the problem. It had been amplifying one that already existed in how the account measured itself.
The founder's instinct — keep funding the number that looks good — was reasonable given what he could see. The account did not have a scaling problem; it had a measurement problem, and every budget increase since launch had been making that measurement problem more expensive.
What we did — the creative that carried the spend
The losers are here on purpose. A test with only winners was never a test.
The authenticity unboxing, pushed to cold audiences at scale
Winner"Original box, original tags — here's the invoice."
- Format:
- Vertical, 18 seconds, unboxing with supplier paperwork on camera
This creative held its cost per purchase as budget roughly tripled, because the objection it answers — is this a real pair — stays relevant to every new cold audience the widening budget reaches. It became the anchor of the prospecting campaign precisely because it did not decay the way a purely aesthetic ad does.
A single best-seller video, pushed to five times its original budget
Lost"The pair everyone's asking about."
- Format:
- Vertical, 15 seconds, product shots cut to music
The classic scaling mistake, and worth showing exactly because it is common: funding one winner harder rather than adding supply drove frequency up fast, and cost per purchase followed it up within two weeks. The ad had not gotten worse; the same people were simply seeing it too often.
The measurement video, cut for Snapchat's women's audience
Winner"Measure your foot in centimetres before you order."
- Format:
- Vertical, 15 seconds, phone camera, a ruler against a bare foot
The proof-led angle that had worked on Meta transferred cleanly to Snapchat once the audience for the women's line was defined properly, confirming the angle rather than the platform had been doing the work. It let the second platform launch with a proven hook instead of a guess.
A generic "shop now" retargeting ad, run at higher frequency
Lost"Still thinking about it? Shop now."
- Format:
- Vertical, 8 seconds, product carousel, no new information
Fine at low frequency and fast to fatigue once retargeting budget increased alongside prospecting, because it never gave the shopper a new reason to return — it just repeated the ask. It was replaced with retargeting cuts of the winning proof-led creatives, which held up far longer.
National Day budget pre-load with three fresh hooks
Winner"Kuwait colours, limited run before the 25th."
- Format:
- Vertical, 15 seconds each, three hook variants over the same colourway
Because three fresh hooks were shot and approved three weeks ahead, the seasonal budget step-up around National Day scaled cleanly instead of forcing the same two tired ads to absorb the extra spend, which is what had happened the year before.
What we did — the optimizations, in order
Net every return out of the return-on-ad-spend number
We rebuilt the weekly report so platform-reported return is checked against revenue net of confirmed returns and refused cash-on-delivery orders, on a lag that matches the store's actual return window.
Why: Platform ROAS only knows about the sale, not the pair that comes back three weeks later. Scaling on the unadjusted number amplifies a gap that was never real profit, and it is a footwear-specific trap that a lower-return category would not hit nearly as hard.
Consolidate the account into one campaign per job, per platform
We collapsed a scattered set of near-identical ad sets on Meta into one prospecting, one retargeting and one retention campaign, and set up the same structure fresh on Snapchat and TikTok.
Why: The old structure had ad sets competing against each other in the same auction and splitting the platform's own learning. One campaign per job let the budget compound instead of fragmenting across a dozen near-duplicates.
Confirm at least six live creatives before any increase
An audit of the account found two ads carrying most of the budget. We paused further increases until the team shipped four more proof-led creatives and confirmed all six were performing.
Why: Adding budget onto two winners is exactly what drives frequency and cost per purchase up together. The fix is creative supply, not caution alone, and it is the step accounts most often skip on the way to a bigger number.
Raise budget in steps, not doublings
Twenty to thirty percent per step, held five to seven days, with a rollback rule set in advance: two consecutive days above the cost ceiling and the previous step returns.
Why: The brand's habit had been to double the budget on a good day and judge it two days later, which is exactly what had driven frequency into a corner in the first place. A slower, ruled step is what let the account find its ceiling without overshooting it repeatedly.
Add Snapchat only once Meta was stable
Snapchat was introduced specifically for the women's line once frequency and cost per purchase on Meta had held steady for two weeks, with budget moving between platforms by blended efficiency rather than either platform's own reported return.
Why: Adding a second platform before the first stabilises splits creative supply and attention thin across both, and it makes it impossible to tell whether a wobble came from the new platform or the old one. Sequencing it is what let the launch be judged fairly.
Pre-load the National Day and Eid budget
Seasonal budget and three fresh creatives locked in three weeks ahead of each date, with a higher accepted cost per thousand agreed in advance rather than discovered mid-campaign.
Why: Waiting to react to a seasonal cost spike means discovering the ceiling the expensive way, in the exact week the calendar matters most. Kuwait's footwear peaks are predictable to the week, so the budget and the creative should be too.
Name the ceiling when it appears
When the net-of-returns return kept falling at every budget step regardless of new creative, we said so directly and redirected the next quarter's growth conversation toward average order value and repeat rate instead of further spend.
Why: Continuing to feed frequency past that point is buying the same customer twice, at a steadily worse price. The honest move is to say the account has hit its ceiling on spend and point the next block of growth somewhere spend cannot reach.
What changed
The table above carries the headline figures, and blended marketing efficiency net of returns is the one worth reading first, because it is the number the platform-reported return had been quietly overstating. It moved because the gap between gross and net closed, not because the platform's own metric changed its story.
Frequency came down and held, and cost per purchase followed it rather than the other way around, which is the order you want to see: a structural fix showing up in the behaviour metric before it shows up in the cost metric. The best-seller-at-five-times ad is worth naming precisely because it is the mistake every account makes on the way up, and undoing it — spreading the same budget across six proven creatives instead of two — did more for the ceiling than any single new hook.
Snapchat launched with a proven angle instead of a guess and reached profitability faster than Meta had in its own first month, which is the return you get from sequencing a platform addition properly rather than adding it out of habit.
What we would do next
Hand the ceiling finding to a retention push. If spend cannot lift return any further without a bigger creative or product-line change, the next block of growth is a second order from an existing buyer, which costs a fraction of a new one.
Test TikTok next, specifically for the streetwear line's discovery traffic, but only after the Snapchat launch has held for a full budget cycle — the same sequencing discipline that made Snapchat's launch clean rather than a scramble.