A brand comes to Converzion Catalyst showing a 3.2x ROAS in Meta Ads Manager. The founder is confused: the numbers look strong, but the bank account tells a different story. Margins are thin. Cash flow is tight. The ad account says they're printing money. Their accountant says otherwise.

This disconnect is not unusual. It's the norm. Platform-reported ROAS is one of the most misleading metrics in performance marketing, and the brands that scale successfully are the ones that learn to stop trusting it and start calculating what actually matters: true funnel ROAS.

The Three Lies Inside Platform ROAS

Meta's Ads Manager reports ROAS using a simple formula: purchase conversion value divided by ad spend. That formula is correct. The problem is that every input feeding it is distorted.

Lie #1: Attribution Window Inflation

Meta's default attribution window is 7-day click, 1-day view. That means if someone sees your ad on Monday, doesn't click it, and then buys your product on Tuesday through a Google search, Meta claims that conversion. If someone clicks your ad on a Tuesday and buys the following Monday — six days later, after seeing a TikTok ad, reading a blog post, and getting an email — Meta still takes full credit.

The 1-day view attribution is particularly aggressive. A user scrolls past your ad in their feed for 1 second without stopping. They don't click. They don't engage. But if they purchase within 24 hours through any channel, Meta reports it as an ad-driven conversion.

In Converzion Catalyst's analysis across 27 eCommerce ad accounts, switching from 7-day click / 1-day view to 7-day click only reduced reported ROAS by an average of 22%. Switching further to 1-day click only reduced it by 41%. Neither change affected actual revenue. The same number of orders came in — they were just no longer being double- and triple-counted across platforms.

22%
Average ROAS drop when removing view-through attribution
41%
Average ROAS drop under 1-day click-only attribution

Lie #2: No Accounting for Returns and Refunds

Meta records revenue at the point of purchase. It never subtracts returns. For fashion and apparel brands running return rates of 20–40%, this isn't a rounding error — it's a structural misrepresentation of actual revenue.

Consider a brand generating £100,000 in Meta-attributed revenue per month with a 28% return rate. Platform ROAS shows £100K in value. Actual retained revenue is £72K. That alone drops a 3.2x ROAS to 2.3x — before accounting for any other costs.

Worse, return rates vary by acquisition channel and creative type. Impulse-driven creatives (flash sales, urgency hooks, "last chance" messaging) consistently produce higher return rates than value-based or education-led creative approaches. A campaign showing 4x ROAS on impulse creative might actually underperform a 2.5x campaign on educational creative once returns are factored in.

Lie #3: Blended vs. Incremental Attribution

This is the most insidious distortion. Platform ROAS treats every attributed conversion as if the ad caused it. But some of those customers would have bought anyway — through organic search, direct navigation, email, or brand recall from previous marketing. The ad was the last touch, not the cause.

The difference between blended ROAS (all attributed conversions) and incremental ROAS (only conversions that wouldn't have happened without the ad) is often 30–50%. For mature brands with strong organic traffic, the gap can be even wider.

This is why a brand can increase Meta spend by 40% and see Meta-reported revenue increase by 40%, while total business revenue only increases by 15%. The platform claims credit for conversions it didn't create.

Building a True ROAS Model

True funnel ROAS requires expanding the calculation in two directions: capturing the full cost stack (not just ad spend) and measuring actual retained revenue (not platform-reported conversions).

Step 1: Define the Actual Funnel

Most brands calculate ROAS as if the funnel is "ad → purchase." The actual funnel for a Meta Ads campaign looks more like this:

  1. Ad impression/click — cost: media spend
  2. Landing page — cost: design, development, hosting, testing
  3. Lead capture or add-to-cart — cost: CRM/ESP platform fees, pop-up tools
  4. Conversion (purchase or sign-up) — cost: payment processing, platform fees
  5. Post-purchase (delivery, returns, support) — cost: fulfilment, returns processing, customer service
  6. Retention (repeat purchase) — cost: email/SMS marketing, loyalty programme

Every stage has costs that the platform-reported ROAS ignores entirely.

Step 2: Assign Real Costs at Each Stage

The cost stack for a typical eCommerce brand running Meta Ads extends well beyond the ad spend line item. Here's what a realistic monthly cost allocation looks like:

For a brand spending £40,000/month on Meta Ads, the true cost of running that channel — including all supporting infrastructure — is typically £50,000–£60,000/month. That's a 25–50% increase over the number being used in the ROAS denominator.

Step 3: Calculate Retained Revenue

On the revenue side, the adjustments are equally significant:

Worked Example: The £40K/Month Brand

Here's a real scenario, anonymised from a Converzion Catalyst client engagement:

Platform-Reported Numbers

Monthly Meta ad spend: £40,000
Meta-reported purchase value: £128,000
Platform ROAS: 3.2x

The Meta dashboard shows a healthy, scalable campaign. The brand's media buyer recommends increasing spend by 30%.

Now apply the true cost model:

True Funnel ROAS Calculation

True cost stack:
Media spend: £40,000
Creative production: £5,500
Landing pages & testing tools: £1,200
Tech stack (Shopify, Klaviyo, analytics): £2,100
Agency management: £4,000
Payment processing (2.1% of retained): £1,895
Total true cost: £54,695

Retained revenue:
Gross: £128,000
Less returns (28%): -£35,840
Less chargebacks (1.5%): -£1,920
Retained: £90,240

True funnel ROAS: £90,240 ÷ £54,695 = 1.65x

That's a 48% reduction from the reported 3.2x. The campaign isn't disastrously unprofitable — but it's not the growth engine the platform dashboard suggested, and scaling spend by 30% without improving unit economics would compress margins further.

3.2x
Platform-reported ROAS
1.65x
True funnel ROAS after all costs

Incrementality Testing: Measuring What the Ads Actually Caused

True funnel ROAS fixes the cost and revenue side of the equation. Incrementality testing fixes the attribution side — answering the question: how many of these conversions would have happened without the ads?

Three methods, ranked by rigour and complexity:

Method 1: Holdout Groups (Conversion Lift Studies)

Meta offers built-in Conversion Lift studies through its Experiments tool. The platform randomly splits your target audience into a test group (sees your ads) and a holdout group (doesn't see your ads but would have been eligible). By comparing conversion rates between the two groups, you get a direct measurement of incremental lift.

The limitation: these studies require significant spend (typically £10,000+ per test period) and 2–4 weeks of runtime to reach statistical significance. They also only measure within the Meta ecosystem — they don't capture cross-channel cannibalisation.

Method 2: Geo-Lift Tests

More robust than platform-native tools. Select matched geographic regions (similar demographics, purchase history, brand awareness). Run ads in one region, go dark in the matched control region. Compare total revenue (not platform-attributed revenue) between the two regions.

This approach measures true business impact rather than platform-claimed conversions. Google's open-source CausalImpact package and Meta's own GeoLift tool both provide statistical frameworks for this analysis.

A Converzion Catalyst client running a geo-lift test across UK postal regions found that Meta was over-reporting incremental revenue by 34%. The ads were still profitable on an incremental basis — but the true incremental ROAS was 1.4x, not the 2.8x shown in the dashboard.

Method 3: Matched Market Studies

The most rigorous approach. Identify markets with similar characteristics (population, income, brand awareness, historical conversion rates), run different spend levels in each market, and measure total business outcomes. This method accounts for every cost and every channel interaction, giving the clearest picture of advertising's true contribution.

The tradeoff: matched market studies require 8–12 weeks, significant operational coordination, and statistical modelling expertise. They're most appropriate for brands spending £20,000+ per month who need to make major scaling decisions.

The Scale vs. Optimise vs. Cut Framework

Once true funnel ROAS and incrementality are measured, every campaign falls into one of three buckets. The decision framework is straightforward — but it requires knowing your margins.

First, calculate your break-even ROAS. If your gross margin (after COGS, shipping, returns) is 60%, your break-even true funnel ROAS is 1 ÷ 0.60 = 1.67x. Any campaign above that threshold generates profit. Anything below it loses money on first purchase.

Scale (True Funnel ROAS > 2x Break-Even)

If break-even is 1.67x and the campaign is delivering 3.0x+ on a true funnel basis, there's clear headroom. Increase daily budgets by 20–30% per week and monitor for diminishing returns. Watch the marginal ROAS on incremental spend — if adding £10K in weekly spend only generates £15K in incremental retained revenue, the campaign is approaching its ceiling.

Optimise (True Funnel ROAS Between 1x and 2x Break-Even)

This is where most campaigns land. Profitable, but not with enough margin to scale aggressively without risk. The priority here is improving unit economics before increasing spend:

Converzion Catalyst's typical engagement with a brand in this zone focuses on improving the conversion infrastructure before touching the ad spend — because a 20% improvement in landing page conversion rate is equivalent to a 20% reduction in effective CPA, which directly increases true funnel ROAS.

Cut (True Funnel ROAS Below Break-Even with No LTV Upside)

If a campaign is below break-even on a true funnel basis and the acquired customers show no meaningful repeat purchase behaviour within 90 days, the campaign is destroying value. Cut spend or reallocate to better-performing segments.

The caveat: if LTV analysis shows that acquired customers have strong retention (repeat purchase rate above 25% within 6 months), it may be rational to accept a below-break-even first-purchase ROAS. That's an LTV-first acquisition strategy, covered in depth in the LTV-First Acquisition framework.

What to Do Next

Pull your last 90 days of Meta Ads data. Map every cost that supports the channel — creative, landing pages, tools, people. Pull your actual retained revenue from your accounting system, not from the ad platform. Divide retained revenue by total cost. That number is your true funnel ROAS.

If it's materially different from what Meta is reporting — and it almost certainly will be — you've just identified the gap between perceived performance and actual performance. That gap is where the real optimisation work begins.

The brands that grow profitably aren't the ones chasing platform-reported ROAS targets. They're the ones that know their real numbers, at every stage of the funnel, and make scaling decisions based on retained revenue and true costs. Everything else is a vanity metric dressed up in a dashboard.

← Previous Article The 7 Checkout Friction Points Next Article → Shopify Performance Optimisation