We showed Coolenator its real conversion rate was 3%, not the 1% its dashboard reported.

Coolenator sells an aluminium wine cooler that takes a frozen bullet and actively chills, where the premium alternatives passively insulate. Performance was not the complaint. Scale was. The account had run EUR 7,977 and 174 purchases in the trailing 90 days under the client’s own management, ROAS sat at 7 to 9x, and it crashed every time budget went up, so spend stayed self-capped near EUR 16,000 a month. The in-house marketing manager had already been moved off DTC onto B2B sales.

Revenue also depended on discounting: “the whole strategy, of course, before that, we always had the discount and that was what pushed people to sell.” And the creative the brand found easiest to make was the creative that lost money. Founder-led ads had returned 0.57x and 0.86x, and a sports tie-in had spent EUR 241 for zero purchases. The account was roughly 95% video, so statics were untested.

Underneath all of it sat a number nobody had questioned. Shopify’s dashboard reported 1.07% for August. On a EUR 59.85 average order, a 1% conversion rate ends an ad ramp before it starts, because it makes every plausible acquisition cost look unaffordable. The client believed it. So did the account manager.

Step #1. Build the account off its own history rather than off opinion.

Creative angles were ranked on the client’s own EUR 7,977 of spend. Anti-ice framing had the best click-through rate in the account at 1.80% to 1.94%, so it led, and founder-led and sports tie-ins were excluded on the evidence rather than on taste. A claims ledger written against the live site fixed what the ads could say. Breakeven was modelled per country off the client’s real shipping and landed costs: 1.75x in the Netherlands at best, 2.24x in Sweden at worst, 2.02x to 3.43x once the discount applies. That spread is tight enough to argue for one campaign instead of nine, because EUR 150 a day split nine ways starves every lane below the volume Meta needs to leave learning. Meta went live on 25 August at EUR 150 a day across prospecting, retention and catalog retargeting.

Step #2. Go and look outside the ad account.

The conversion rate did not match the funnel behind it, so we pulled Shopify’s sessions dataset live and computed the rate ourselves, as sessions that completed checkout divided by sessions. Shopify’s own bot classification accounted for 2,901 sessions over 90 days, 16.3% of all traffic, producing five checkouts. Another 1,784 human sessions came from countries the store has never sold to, mostly the US at 885, producing zero. That is 4,685 of 17,841 sessions sitting in the denominator of every rate the client reads. The August spike traced to one week: 969 bot sessions in the week of 10 August, 906 claiming Chrome on Mac OS X and 12 announcing HeadlessChrome. One spoofed scrape, farming the 10% email signup the client had re-enabled.