How measuring the real per-market break-even from 10,207 order-level records, then setting every budget from marginal return instead of platform ROAS, took store revenue from £2,015,906 to £3,602,841 in twelve months and made the biggest scale test return 3.00x.
Key results.
CHUUG grew 78.7% across the first full twelve months after the takeover, measured on store data, not platform attribution.
Challenge
CHUUG arrived with a good product and a founder whose stated blind spot was marketing. The previous Meta agency was paid 3.5% of net profit and, in his words, had incentives that were never aligned. Meta ran at roughly 3.4x and was not being scaled or proactively managed. Google had been switched off entirely during 2025 after a run at roughly 1.x ROAS, and our rebuild-readiness audit scored what was left at 18%. The brand was undefended in search: no Brand Search campaign existed, the brand term was negated account-wide, and Shopping branded impression share sat at 58.4%.
The deeper problem was that nobody knew the numbers the business was run on. The break-even ROAS in the system was 1.74x, authored and never measured, so budget was allocated on gross margin percentage instead. And the real ceiling was never marketing. It was the factory.
Solution
Step #1 Measure the economics before touching the budget.
We derived the break-even from 10,207 order-level records rather than from an assumption: 1.61x blended, retiring the 1.74x. When the client re-costed every shipping zone to carrier actuals in August 2026, we re-derived the stack within hours on 5,594 orders: UK 1.48x, Canada 1.64x, Australia 1.71x, US 2.04x, 1.83x blended. AOV measured at £96.89 and LTV at about 1.05 times AOV, which killed any scaling case funded on repeat purchase. Units per order measured at 2.040, and the 1.8 in use had been over-buying orders by about 13%.
Step #2 Rebuild both channels against those numbers.
On Meta a 109-point audit scored the account at 74%: eight campaigns against a doctrine maximum of three, and 160 of 164 ad sets starved below the learning threshold. Of 824 active ads, 286 had spent under £5 in 30 days, held in place by a hard Page cap of about 1,000 ads that quietly stops new ads delivering; pausing the 197 oldest freed about a fifth of it. On Google we rebuilt from the dead account and built Brand Search for the first time in its life. Thirty days after relaunch it returned 3.33x on £20,213 of spend, and monthly Google spend went from £3,288 in January 2026 to £31,171 in August.
Step #3 Set every budget from marginal return, not reported ROAS.
Google’s headline account ROAS reads 5.5 to 6.5x. We showed that £1.1k of brand spend had “earned” £98k at 50 to 100x, capped brand at about £25 a day and stripped it out of every budget decision, leaving real incremental Google at 2.9 to 3.4x, worse than Meta. We then measured marginal return across 64 material spend moves in the account’s history, store revenue by shipping country against spend by country. Store-measured marginal came back at 2.44x against Meta’s reported 2.48x, correlation 0.87. So when the next UK pound was returning about 0.5x against a 1.48x break-even, we recommended trimming the client’s best-margin market.
Step #4 Work outside the ad account.
A site audit across all six live markets produced 39 findings, 34 already built into a draft theme before delivery: a builder at 88% scroll depth on a page taking roughly £60,000 a month of Meta traffic, and a checkout taking AED 662 where that builder promised AED 566.
Results
Book a discovery call about your business goals
- ✓Non-binding introductory meeting
- ✓Explore your growth potentials
- ✓Talk to our experts