Every audit we run starts with the same question: what would happen to revenue if you switched this campaign off tomorrow? Nine times out of ten, nobody knows. Platform-reported ROAS answers a different question — it tells you which ad got the last touch, not which spend created demand that did not already exist.
Split comparable regions, suppress spend in one, and compare total revenue. It is the cleanest signal available to most mid-market advertisers and needs no engineering work. Run it for at least four weeks so seasonality does not distort the read.
Reduce brand search spend in controlled increments and watch organic click recovery. Many advertisers discover they can reclaim twenty to forty percent of brand budget with negligible revenue loss, then redeploy it into prospecting where it actually creates new customers.
Keep one audience on a fixed control creative while everything else rotates. This isolates creative lift from bidding noise, which is where most of the real performance variance lives.
Server-side tagging, consent-safe conversion APIs and a warehouse copy of your conversion data are no longer optional. Without them, smart bidding optimises on a partial picture and reports on an even more partial one. Fix measurement first; every budget decision after that gets easier.
Incrementality is not an academic exercise. It is the difference between a media budget that scales the business and one that simply photographs it.
Once measurement is honest, the biggest remaining lever is creative volume. We aim for a minimum of eight distinct concepts per quarter per core audience, each with a clear hypothesis: a different pain point, proof mechanic or format. Accounts with disciplined creative pipelines routinely beat identically budgeted accounts by thirty percent or more.
Do this once and you will never argue about attribution dashboards the same way again.
Chat with our paid team on WhatsApp.