We got Moonster Leather’s branded search converting at 5.7%.

The brand was founded on Amazon and never left. It grew to almost $8M of topline there and spent $550,000 on Amazon advertising in the trailing twelve months at a reported ROAS of around 5.7. Outside that marketplace it barely existed. Its own Shopify store sold roughly 20 units a month, which the CEO called irrelevant, and site conversion rate sat under 2%. External ad spend across the brand’s whole life was zero.

The incumbent social agency of six or seven years produced content the CEO described plainly: “I get a lot of creatives from the Instagram partnerships, but they don’t convert, right? So they don’t convert at all.” And Meta, the channel the plan was built for, was locked: the ad account and every social asset had gone to a German company that went insolvent, and the last known owner was unreachable.

Step #1 Build the channel from zero, four markets, over Black Friday.

We staggered the launch by market rather than fighting peak-day auctions everywhere at once: US on the Friday, UK next on recycled US creative, Germany and France by the Thursday. It went in hyper-segmented at roughly 30 ad groups so budget could move off early signal fast, with branded search, non-branded search and Performance Max as separate lanes behind GA4 and Merchant Center tracking. Scaling gates were set before launch: scale at 2.86 ROAS, no cap above 3.4.

Step #2 Call the launch wrong in its first weekend.

The first weekend review found Performance Max burning roughly 60% of budget on poor-quality display, because it was learning from Google’s auto-generated assets rather than the real feed. Journals were showing in a “Leather Bags for Men” ad group, and the copy promised custom and personalised products the brand does not make. We stripped the mismatched assets out the same day and stood up standard Shopping.

Step #3 Read the lanes and say which demand was real.

On December 9 we went through the account with the founder. Branded search was the healthy lane at 5.7%, so its budget went from GBP 15 a day to GBP 20. Non-branded had produced 4 conversions at a CPA above GBP 100, unsurvivable against a 14-day AOV of $41.20, so it was cut back to a target CPA and run as awareness. The reading that mattered was not about media. Volume rose while the 7-day average order value fell 29% to $34.87, because the mix skewed to low-value journals rather than the bags the plan depended on. The work moved to basket size.

Step #4 Refuse the fast route on Meta.

Meta could have started immediately through the incumbent agency’s partner access. We said no: the spend would have sat inside an account the client could not recover if it were restricted, and a from-scratch account would have thrown away years of pages, followers and pixel history. We chose the formal ownership transfer instead. It never completed.