The problem with the ROAS Meta or Google shows you
Open your Meta Ads dashboard. Your ROAS is 4.5×. Great. That means for every pound you invest, you're generating 4.5 in sales — right? Not exactly.
What Meta shows you is attributed ROAS: the sales the platform claims credit for generating, based on its own attribution model. That number includes conversions that would have happened anyway, conversions counted multiple times across different platforms, and conversions attributed even when the customer had already decided to buy before seeing your ad.
The result is that the ROAS shown in platform dashboards tends to be systematically inflated — on average 40% to 200% above the real ROAS, depending on the sector and campaign type.
An advertiser optimising for an apparent ROAS of 4.5 could actually be generating a real ROAS of 1.8×. With a 40% margin, that means they're losing money on every pound spent on advertising — and the platform's data never tells them.
The 4 causes of the gap between apparent and real ROAS
1. Multi-touch attribution problems
A customer sees your ad on Meta on Monday. On Tuesday they search for your brand on Google and click a search ad. On Wednesday they get an email from you and buy. How many platforms will claim credit for that sale? All three. The same conversion gets counted three times across three different dashboards. Your total ROAS is therefore 3× inflated relative to the reality of that single transaction.
2. Organic conversions attributed to advertising
A customer who's followed your brand for months, already bought twice and searched for you directly on Google — but happened to scroll past one of your display ads without even reading it — gets counted as an advertising conversion. Google Ads claims credit for a sale that would almost certainly have happened without that ad. This phenomenon is common enough to have a name: ad attribution cannibalisation.
3. Extended attribution windows
By default, Meta attributes conversions that happen within 7 days of a click and 1 day of an ad view. That means if someone sees your ad on Monday and buys the following week for a completely unrelated reason, that sale gets attributed to your campaign. Extended attribution windows systematically inflate the numbers, especially for products with long purchase cycles.
4. Returns and refunds aren't subtracted
The platform counts the sale the moment it happens. If the customer returns the product two weeks later, the platform doesn't know. Your apparent ROAS includes all gross sales — including the ones that turn into returns, chargebacks or refunds. For sectors with high return rates (fashion, electronics), this can distort ROAS by 20-35%.
How to calculate real ROAS: the correct method
There are two approaches to measuring ROAS accurately. The first, more precise but more complex, is the incrementality test (or holdout test). The second, more accessible, is the MER — Marketing Efficiency Ratio.
The Marketing Efficiency Ratio (MER)
MER is the most honest metric you can calculate without sophisticated tools. The formula is simple:
Instead of looking at ROAS campaign by campaign the way the platforms show it, you look at the ratio between the total sales generated by your business and the total ad spend. This number is immune to multi-touch attribution problems because it doesn't try to attribute every single sale to a specific source.
If last month you generated €120,000 in total revenue and spent €30,000 on advertising (Meta + Google + TikTok), your MER is 4.0×. That's the real number to base budget decisions on — not the 6.2x Meta shows you or the 5.8x Google shows you.
Margin-adjusted ROAS
MER measures revenue against ad spend. But revenue isn't your profit. To really understand whether your campaigns are working, you need to work with margin-adjusted ROAS:
If your net margin is 30%, your break-even ROAS is 1/0.30 = 3.33x. Any real ROAS above 3.33x generates profit. Any real ROAS below it generates losses. With this formula, an apparent ROAS of 4.5x that turns out to be 2.8x in reality is actually a campaign eroding your margin with every pound spent.
The two scenarios that change everything
Meta shows you 5.2x. You're happy. You increase the budget. But the real MER is 2.1x. With a 35% margin, you're actually losing £0.14 for every pound spent on advertising. The more you spend, the more you lose.
Your MER is 3.8x. With a 35% margin, you're generating £0.33 in profit for every pound spent on advertising. You know exactly how far you can scale before hitting your target break-even ROAS.
The metrics that actually matter in 2025
Here's the metrics dashboard we recommend to our clients, ranked by importance:
| Metric | What it measures | Review frequency | Value |
|---|---|---|---|
| MER (Marketing Efficiency Ratio) | Total revenue / total spend | Weekly | Priority 1 |
| nCAC (new Customer Acquisition Cost) | Cost to acquire a new customer | Monthly | Priority 2 |
| LTV:CAC ratio | Lifetime value vs acquisition cost | Monthly | Priority 3 |
| CPM / CPC per creative | Efficiency of individual creatives | Daily | Operational |
| Platform ROAS (raw) | Platform-side attribution | Directional proxy only | Don't optimise for this |
How to set up accurate tracking
To get more accurate data, you need to implement a tracking system that narrows the gap between the platform and reality. Here are the fundamental steps:
Implement server-side tracking
Server-side tracking reduces data loss caused by ad blockers and iOS 14+ by up to 80%. It's the foundation of any accurate measurement. It requires a developer to implement, but it's essential.
Use shorter attribution windows
Set Meta and Google to 1-day click / 0-day view attribution windows for products with a short purchase cycle. You'll lose credit for some genuine conversions, but the data will be much cleaner.
Separate new customers from existing customers
ROAS on existing customers is always higher, but it doesn't reflect your ability to acquire new ones. Track the acquisition cost of new customers separately to really understand the health of your growth funnel.
Run periodic holdout tests
At least once a quarter, turn off a channel for 2 weeks on a controlled audience segment and measure the impact on real revenue. This tells you the true incrementality of that channel.
Subtract returns and refunds from your metric
Connecting your order management system to your dashboards lets you calculate ROAS on net revenue, not gross. For sectors with return rates above 10%, this correction is essential.
The practical implications: how campaign management changes
Once you start optimising for real MER instead of apparent ROAS, some decisions that seemed obvious turn out to be wrong — and vice versa.
Case 1: The "weak" campaign that was actually working. A brand awareness campaign with an apparent ROAS of 1.8x looks like a disaster. But its real role is to lower the CPM on conversion campaigns (through brand recall) and increase email conversion rates. Measuring MER before and after switching this campaign off often reveals it was contributing positively to the overall system's efficiency.
Case 2: The "excellent" campaign that was just cannibalising. A retargeting campaign on already high-intent customers shows an 8x ROAS. It looks like the best possible investment. But if you switch that campaign off, those customers buy anyway — simply through another channel. The real incrementality could be close to zero, with the apparent ROAS entirely illusory.
Use platform metrics (CPC, CPM, frequency, CTR) to optimise individual creatives and campaign structure. Use MER to decide how much budget to allocate to each channel. Don't mix the two levels of analysis.
Conclusion: the right number changes everything
The difference between a business that scales its advertising profitably and one that burns budget by ramping up campaigns often comes down to this: knowing which number to optimise for.
Platform-reported ROAS is a useful tool for understanding relative trends — whether one creative outperforms another, whether one audience responds better. But it's not the tool for deciding whether to increase or decrease overall spend, and it's not the number to use when evaluating the real impact of your advertising on the business.
MER is imperfect, like every metric. But it's honest. And honest data you can make imperfect decisions on is always better than distorted data that produces perfect decisions for the wrong metric.
Want an analysis of your real ROAS?
We'll show you the gap between the ROAS you're looking at and what you're actually generating — and how to optimise your advertising system to maximise real profit.
Analyse my campaigns