ROAS tells you which channels reported the sale, not which ones caused it.
Key takeaways
- Going beyond ROAS means measuring what your marketing caused, not just what it recorded. Return on ad spend credits the last click and misses the demand that was already there.
- ROAS flatters channels that would have converted anyway, especially branded search and retargeting, so a high number can hide a low true return.
- Incrementality answers the only question that matters: what changed because you spent. It is the check a strong ROAS still needs.
- The marketing that compounds (brand, retention, category demand) often shows a weak ROAS because its payoff lands later and off-channel.
- A good measurement system shows what is working and why, blending attribution, incrementality tests, and the new AI-era signals of visibility.
What does "beyond ROAS" actually mean?
Going beyond ROAS means building a measurement system that shows what your marketing caused, not just what it recorded. Return on ad spend divides revenue by ad cost inside a channel, which credits whoever reported the conversion last. It is a useful efficiency ratio and a poor causation test. Beyond ROAS is the move from "what got the click" to "what changed the outcome."
The distinction is not academic. Budgets get moved on ROAS, and moving budget toward the channel with the highest reported return often means moving it toward the demand you already had. You optimize into your own base, watch the ratio hold, and wonder why total growth stalled. The number looked healthy the whole time.
Alive Method, a marketing and advertising company, builds measurement so a team can see the difference between recorded revenue and caused revenue, and spend against the second one.
Why does ROAS flatter the wrong channels?
ROAS flatters channels that harvest demand rather than create it. Branded search, retargeting, and email to existing customers all report strong returns because they capture people already on their way to buying. The channel takes credit for a conversion it mostly witnessed. The result is a high ROAS attached to low incremental value.
Consider the pattern. A shopper sees a brand campaign, decides to buy, searches the brand name, clicks the branded ad, and converts. Branded search books the ROAS. The brand campaign that created the intent shows almost nothing, because its effect landed on a different channel days later. Reward the channel with the visible ratio and you defund the one that actually moved the shopper.
This is the core of the ROAS limitations problem: the metric is measured per channel, but buying does not happen per channel. It happens across touchpoints, over time, with most of the influence invisible to any single platform's report.
What is incrementality, and why does it matter?
Incrementality is the measure of what happened because of your marketing that would not have happened otherwise. You establish it by comparing a group exposed to marketing against a comparable group that was not, through geo tests, holdouts, or controlled experiments. It matters because it is the only method that isolates cause from coincidence.
A strong ROAS and a strong incrementality read are not the same claim. A campaign can post an excellent return and still be largely harvesting demand that existed already. Incrementality is the check that tells you whether the spend built something or simply stood next to a sale. High reported returns still need that check behind them before you scale.
| Signal | What it tells you | What it hides |
|---|---|---|
| ROAS | Efficiency of reported conversions in-channel | Whether the sale would have happened anyway |
| Attribution | Which touchpoints were involved on the path | True cause across time and off-channel effects |
| Incrementality | What the spend actually caused | Nothing, when the test is designed well |
| AI-era visibility | Whether you appear in AI answers and search | Direct revenue, which needs to be modeled separately |
What are the new AI-era signals to measure?
The new signals are about visibility inside AI answer engines and search, where a growing share of discovery happens before any click. Whether your brand is cited in an AI answer, surfaced in an AI Overview, or recommended in a generative response is now a demand-creation signal that classic attribution cannot see, because there is often no click to attribute.
These signals behave like brand and category demand: they compound, they show up off-channel, and they resist last-click credit. That makes them a natural fit for a beyond-ROAS approach. You track presence and share of answer over time, then use incrementality tests to connect that visibility to lift, rather than expecting a clean per-click return the medium does not produce.
How Alive Method approaches this
Method treats measurement as a system, not a dashboard, and it lives in the Evolve stage of the Alive Method System: the practice of reading what happened, understanding what it means, and deciding what to do next. We combine attribution, incrementality testing, and the AI-era visibility signals so a client can see both the return and the reason behind it.
Two client results show the two halves of this. Montis Pickleball ran roughly 7 to 8x ROAS on Google Ads (reported figure). Strong, and exactly the kind of number that still needs incrementality behind it before you assume all of it is caused rather than captured. The Tradition posted 693 conversions in a quarter with cost per conversion down about 20% (reported figure): measured improvement against a baseline, which is the shape of progress you can actually trust. One is a headline number. The other is a proof. The work is knowing which is which.
FAQ
Is ROAS a bad metric?
No. ROAS is a fine efficiency ratio and a quick health check on a channel. It becomes a problem when it is used alone to judge whether marketing is working or to decide where budget should move, because it cannot separate demand you created from demand you captured.
What is a good ROAS?
There is no universal number, because it depends on margin, price, and how much of the reported return is incremental. A 7x ROAS on a channel that mostly harvests existing demand can be worth less than a 3x on a channel that creates new customers. Judge the return against incrementality, not against a benchmark.
How is incrementality measured?
Through controlled comparison: geo holdout tests, audience holdouts, or matched-market experiments where one group is exposed to marketing and a comparable group is not. The gap between them is the incremental effect. It takes design and patience, which is why it is run as a program, not a one-off report.
What is "marketing that compounds"?
Marketing whose payoff builds over time rather than converting immediately: brand, retention, category demand, and now AI-era visibility. It usually shows a weak ROAS because its effect lands later and off the channel that spent the money. Measured only by return on ad spend, it looks like a cost. Measured by incrementality and long-term value, it is often the growth.
Can attribution and incrementality work together?
Yes, and they should. Attribution maps the path and helps with day-to-day optimization. Incrementality validates whether the spend caused the outcome. Attribution tells you what happened; incrementality tells you what to believe about it.
Where to start
A high ROAS is a fine thing to have and a dangerous thing to trust on its own. The teams that grow past a plateau are usually the ones that stopped asking which channel reported the sale and started asking which spend caused it. That is a measurement system, and it is buildable.
Tell us what you're trying to achieve. Start the conversation.