Beyond ROAS: A Modern Playbook for Measuring What Actually Moves Growth
- Marcio Fasano
- Oct 10
- 2 min read
If you still judge marketing by last-click ROAS and a neat linear funnel, you’re flying by old instruments. Privacy changes, walled gardens, and messy real-world journeys have made “ROAS + funnel” a narrow—and often misleading—view. Here’s a practical, research-backed way to measure incremental, profitable growth in 2025.
Why ROAS (alone) fails
ROAS often rewards easy conversions, ignores baseline sales you’d have gotten anyway, and under-credits channels that work higher in the journey (brand, video, social). It also misses cross-channel effects and lifetime value differences—so you optimize for the wrong thing.
Upgrade the target:
POAS (Profit on Ad Spend) or contribution margin, not just revenue—optimize to profit, not turnover.
Incrementality (what changed because of ads), not just attributed conversions.
The new measurement stack
1) Causal experiments (prove lift)
Geo holdouts/geo-experiments: split regions into test vs. control to estimate causal impact at market scale; privacy-safe and channel-agnostic.
Platform lift tests: e.g., Meta Conversion Lift for on-platform incrementality.
Ghost ads/bidding: control-exposed methods adapted for the open web to isolate incremental effect.
Caveat: geo tests can be costly/complex to run well; plan design, timelines, and power in advance. Recast
2) Model the big picture (MMM)
Run Marketing Mix Modeling to quantify channel contributions over time and scenario-plan budgets (including brand and offline). Open-source options like Robyn and LightweightMMM make MMM accessible.
3) Embrace privacy-centric analytics
Prepare for a world with patchy user-level data: GA4 modeling (behavioural & conversion) and Consent Mode fill some gaps when consent is missing, but you must interpret modeled data carefully.
4) Measure attention, not just viewability
Attention metrics (e.g., Adelaide’s AU) aim to predict outcome-driving media quality. Research suggests attention can outperform viewability at predicting results—use with creative and frequency controls.
What to optimize to
POAS / Contribution margin (profit), not just ROAS.
Incremental CAC and Incremental Revenue from lift tests.
mROAS (marginal ROAS) from MMM or experiments to find the next best dollar.
LTV-based targets and payback windows for durable growth. (Synthesis across sources.)
Quality of exposure via attention and creative diagnostics.
A simple operating system for modern measurement
Quarterly:
Refresh MMM (Robyn/LightweightMMM). Run at least one geo or platform lift test on a big channel. Prioritize budgets by mROAS and profit.
Monthly:
Validate modeled GA4 trends vs. transaction systems; monitor consent rates and modeled-data share.
Rotate creatives toward high-attention placements; cap wasted impressions.
Weekly:
Guardrails: blended POAS, incrementality sanity checks (ghost/holdout spot tests), and fatigue/overlap reviews.
Join the conversation
What’s been your biggest measurement win, or blind spot, in the last year?
Share your experience, challenge the ideas above, or get in touch if you want a one-page roadmap (with test design and POAS targets) tailored to your stack.
Advertisers: Join Here Publishers: Join Here
Comments