Return on ad spend is the most widely used performance metric in paid social, and one of the most misleading. That is not an argument for ignoring it. It is an argument for understanding what it actually measures, what it does not, and why optimising towards it in isolation produces worse business outcomes than the number itself suggests.
This matters more now than it did three years ago, because the ROAS figure in your ads manager has become less reliable at the same time that more brands are treating it as their primary decision-making metric.
The Attribution Problem Inside ROAS
ROAS is a ratio: revenue attributed to ads divided by ad spend. The operative word is attributed. As attribution has become less accurate — through iOS privacy changes, third-party cookie loss, and platform modelling — the revenue figure feeding that ratio has become less precise.
Meta’s reported conversions now include a significant proportion of modelled data. The platform estimates conversions it can no longer directly observe, based on probabilistic signals. This modelling is sophisticated and generally directionally accurate, but it is not measurement. It is inference. And it means that two campaigns with the same reported ROAS may have very different actual returns, depending on how much of the reported revenue is real versus modelled.
The consequence is that brands optimising towards platform ROAS targets are, in part, optimising towards a model of their own performance rather than the performance itself. This creates decisions that look rational inside the ads manager and look confusing when you check them against actual revenue.
The Double-Counting Problem
Add together the attributed revenue from all your paid channels and compare the total to your actual revenue. For most businesses running paid social alongside paid search, display, and email, the sum of attributed revenue across channels will substantially exceed total revenue.
Each platform attributes the conversion to itself. A customer who sees a Meta ad on Tuesday, clicks a Google search ad on Thursday, and converts on Friday will appear as a Meta conversion and a Google conversion. Your actual revenue counts it once. Your platform reporting counts it twice.
This is not a new problem, but it is a persistent one. Brands that are making channel investment decisions based on platform-reported ROAS without accounting for overlap are systematically overestimating the contribution of each individual channel.
What ROAS Does Not Tell You
Even setting aside attribution accuracy, ROAS as a standalone metric is missing several pieces of information that matter for actual business performance.
It does not tell you about margin. A 4x ROAS on a product with 20% gross margin is not profitable. A 2x ROAS on a product with 70% gross margin might be excellent. ROAS does not care about the economics of what you are selling.
It does not tell you about customer quality. A campaign generating high ROAS through discounted or promotional offers, or through acquiring customers who buy once and leave, may look strong in the short term and look very different over a 12-month cohort view.
It does not tell you about incrementality. Some of the revenue attributed to your ads would have happened anyway, through organic search, direct traffic, or word of mouth. True incremental ROAS — what actually happened because the ad ran, not what happened in the same period — is almost always lower than reported ROAS.
And it does not aggregate across channels in a way that gives you a useful view of total marketing efficiency.
The Metrics Worth Building Around
None of this means abandoning ROAS entirely. It means contextualising it within a broader measurement framework.
The metrics that tend to give a more complete and more honest view of marketing performance include the following.
Marketing efficiency ratio (MER) is total revenue divided by total ad spend, across all paid channels. It is blunt, but it is honest. It sidesteps inter-channel attribution arguments and gives you a single view of how much revenue your overall paid investment is generating. Tracked over time, it is a reliable signal of whether your paid marketing is becoming more or less efficient as a whole.
New customer acquisition rate and blended customer acquisition cost tell you whether you are actually growing your customer base or simply reactivating existing customers at an inflated apparent return. Many brands discover, when they look at this properly, that their paid social is generating a high proportion of returning customer purchases — which is valuable, but not what acquisition campaigns are supposed to be doing.
LTV:CAC ratio — the ratio of customer lifetime value to customer acquisition cost — is the metric that most directly connects marketing spend to long-term business value. A brand with strong LTV can afford a higher CAC and a lower initial ROAS than a brand with weak retention. Understanding this ratio is what allows you to make sensible decisions about how much you can afford to spend to acquire a customer.
Payback period — how long it takes to recover the cost of acquiring a customer through their subsequent purchases — matters particularly for brands with longer purchase cycles or subscription models. An eight-week payback period at scale is a very different business from a 14-month payback period, regardless of what the platform is reporting as ROAS.
Brand search volume and direct traffic are useful leading indicators that paid activity is generating awareness beyond last-click attribution. Brands that are working tend to see organic search volume growing alongside paid investment. If paid spend is high and brand search is flat, something is worth investigating.
The Incrementality Question
The most rigorous test of whether your paid activity is actually working is incrementality testing: running controlled experiments that measure what would have happened without the ads, and comparing it to what happened with them.
Incrementality tests are more complex to run than standard campaign analysis and they require holding back spend in a way that most media buyers find uncomfortable. But they consistently produce more honest answers than attribution reporting, particularly for brands where a significant proportion of conversions would have happened anyway through organic channels.
The brands doing this well are not abandoning their standard reporting. They are running incrementality tests periodically to calibrate their understanding of how much credit their paid channels actually deserve, and using that calibration to interpret their day-to-day reporting with appropriate adjustment.
The Practical Implication
Most brands are not going to overhaul their entire measurement approach immediately. But there are some practical steps that make the picture more honest without requiring a complete infrastructure rebuild.
Track MER alongside channel-level ROAS, and watch how it moves in relation to total spend. Look at new customer rate within your reported conversions, and understand what proportion of your “acquisitions” are actually returning customers. Build a simple LTV model for your main customer cohorts, even if it is approximate, and use it to set ROAS targets that reflect the actual economics rather than a platform benchmark. Run at least one incrementality test this year to calibrate your attribution assumptions.
None of this is about having less data. It is about having data that is actually telling you something real — and making decisions from a position of genuine understanding rather than misplaced confidence in a metric that was always more complicated than it appeared.
0 Comments