We grew 700 Miles Publishing from $205K a year to $721K a year in sales, annualized.

The account came from a departed agency. We harvested it rather than scorching it, tapering the inherited campaigns while building alongside them. Prospecting was split into five country campaigns (US, UK, Australia, New Zealand, Canada), each starved of volume. The two biggest markets looked weak. The diagnosis was offer, not market: a low-AOV single-item funnel at around EUR 11.

Underneath sat a number nobody had checked. The model justifying any spend at all on a new reader assumed a 22% repeat rate, so every budget decision rested on an assumption doing a measurement’s job. The account also has one attribution source and still does: no Triple Whale, no GA4, no third-party incrementality tool. Nothing here can prove that incremental spend produces incremental customers, and an honest read says so.

Step #1: measure the number the model was guessing. The returning-customer split from Shopify gave a real figure: 46.5% on 679 returning customers, against 22%. A reader coming back at that rate is worth a modelled $205 over twelve months, and at about 38% first-purchase contribution margin that is roughly $78. A EUR 30 acquisition cost is covered about 2.5 times over. The LTV figure is a model and labelled as one wherever it appears. The repeat rate is not.

Step #2: rule the gate metric and publish the derivation. On 11 July we ruled the economics in writing: a Phase-1 cost per purchase band of EUR 12 to 15, a hard ceiling of EUR 30, every gate in the account’s own currency, and first-purchase return explicitly not judged against the 3.4x blended goal used elsewhere. Before that ruling, a good week looked like a bad one.

Step #3: consolidate, then scale in steps against the ceiling. The five country campaigns became one CBO in late June and prospecting rose in steps: EUR 30 a day, EUR 160 at the consolidation, EUR 460 in early July, EUR 490, then EUR 660 on 15 August, each step gated on cost per purchase rather than on return. Michael’s August increase went live the same day he asked, all of it into the two cold-acquisition lanes, verified with two independent read-backs. Every active ad went through the pause tree first and none qualified for pausing. The reversal condition was written down before anything moved: above EUR 15 on seven-day cost per purchase, step prospecting back toward EUR 490.

Step #4: reconcile against the store every week, on counts. The weekly report ties Meta’s purchase count to Shopify’s order count against two independent Shopify backends, and splits new from returning customers. In the week to 2 August store revenue fell 11.5%, which read as the ads breaking. New-customer orders were 442 in that week and 442 in the one before, exactly flat, while returning orders fell 228 to 151. The decline was repeat purchase. Nothing in the account needed changing.