How to use this hub
Use ROAS and funnel math for day-to-day optimization, MER for top-down alignment, marginal ROAS to understand scale limits, and incrementality tests to validate true lift. You don't need all of them at once-add layers as spend grows.
Which metric should you trust-
| Metric | Strength | Blind spots |
|---|---|---|
| ROAS | Fast feedback per campaign/ad set | Can be biased by attribution and window settings |
| MER | Top-down reality check | Hides channel-level performance and creative effects |
| Marginal ROAS | Scale lens (saturation) | Needs clean spend vs incremental revenue data |
| Incrementality lift | Causal truth (holdout) | Harder to run; needs good experiment design |
A pragmatic measurement stack
- Early stage: funnel + ROAS with profit guardrails (break-even ROAS/CPA).
- Scaling: add MER and marginal ROAS to avoid over-crediting retargeting.
- Mature spend: run incrementality tests when you see diminishing returns or channel disputes.
- Always: keep attribution window consistent when comparing periods and channels.
Failure modes to watch
- ROAS looks great while overall growth stalls (likely attribution bias or audience saturation).
- MER worsens while platform ROAS improves (tracking mismatch, spend mix shift, or delayed revenue).
- Marginal ROAS falls as you scale (normal diminishing returns; change creatives/audiences or cap spend).
- Creative fatigues and CTR drops (refresh creative; broaden audiences; rotate placements).