Reading ยท Measurement
Why you should stop optimising to platform ROAS (and what to use instead)
5 min read
Platform ROAS, the number Meta or Google reports, counts conversions each platform takes credit for, often double-counting and over-attributing. Optimising to it makes you scale back the channels that actually drive incremental sales. Instead, anchor on blended MER, total revenue divided by total ad spend, and new-customer CAC against LTV, and use incrementality tests to see what truly moved the needle.
The double-counting problem
If Meta and Google both claim the same sale, your combined reported ROAS can exceed reality. Founders who trust it end up over-crediting retargeting and starving prospecting.
Blended MER is the honest number
Total revenue over total spend, read at the account or business level, cannot be double-counted. Track it weekly and watch the trend as you scale.
Prove incrementality
Geo holdouts, spend-down tests and conversion-lift studies show what sales you would have lost without the spend. That is the number worth optimising to.
Key takeaways
- Platform ROAS double-counts and over-attributes
- Blended MER cannot be gamed the same way
- New-customer CAC versus LTV shows real efficiency
- Use incrementality tests to find what truly drives sales
Frequently asked questions
Is platform ROAS useless?
Not useless, but directional. Use it to compare creatives within a platform, not to judge total business impact.
How do I start measuring incrementality?
Begin simple: a geo holdout or a controlled spend-down on one channel, and compare total revenue against the forecast.