New revenue up 51% in four months. Meta's cost per new payer fell from €77 to €46 while spend rose 51%.

The challenge

Timeleft is a social dining app running paid at scale across Meta and TikTok. When I came in around May, new subscribers were running 40% or more below the previous year, month after month. The Meta account was spending heavily but not reaching new people, and new revenue was flat while the business leaned on repeat customers.

The approach

Stabilise the account before scaling it: a single creative testing campaign with a spend gate and a cost per subscription threshold, winners graduated into every market, regular pause audits to clear fatigued creative, and a weekly reality check comparing Meta spend by country with the client's real new revenue so budget moves toward true return rather than in-platform ROAS.

What I did

• Set up an always-on creative testing campaign with a cost per subscription threshold, and a graduation process into all English-speaking, EMEA, North America and LATAM campaigns.
• Ran pause audits against campaign average cost per action to clear underperforming ads.
• Built a weekly reality check comparing Meta spend per country against cockpit new revenue, and used it to move budget between markets.
• Applied a creative naming taxonomy (persona, angle, hook, format) so creative analytics could show what was working.
• Restructured the North America TikTok app campaigns from subscription to registration optimisation, launched a web registration campaign and built Spark Ads from creator content.
• Set up a new Portugal ad account replicating the main account's retargeting and exclusion structure.
• Produced weekly steering committee notes for the exec team on new revenue, blended cost per acquisition and market movement.

The outcome

• September was the first month of the year with new revenue up on the previous year, after months of 20 to 50% declines.
• Meta's new payers went up 2.5× on 51% more spend, so cost per new payer fell by 40%.
• New revenue rose every month from June to September.

Attribution limitations

Meta's contribution comes from a sign-up survey rather than click tracking, so it is a user-reported attribution. Timeleft's other channels and product changes also ran over the same period, so the growth is not Meta's alone. Figures are from the client's internal reporting. Spend is Meta prospecting and retargeting, averaged over May to September.