Cash ROAS vs Platform ROAS: Which Number Runs Your Business

Two numbers describe the same month of ad spend, and they almost never agree. One lives in Ads Manager. The other lives in your bank account.

Most operators pick the one that feels better and call it a strategy. I would rather you pick the one that can pay your rent, and keep the other for what it is actually good at.

What is the difference between cash ROAS and platform ROAS?

Platform ROAS answers "what does Meta believe my ads caused?" Cash ROAS answers "what did my business actually collect?"

Platform ROAS

It is the conversion value Meta attributes to your campaigns divided by what you spent, under whatever attribution setting the ad set uses. It is fast, available by ad, and the only view that tells you which creative is carrying the account.

Cash ROAS

It is collected cash from customers who came in through ads, minus refunds, chargebacks and failed payments, divided by the same spend. It is slow, it needs a bank or processor export, and it is the only version that can be wrong in your favor or against you without anyone's opinion involved.

I treat them as two instruments, not rivals. One is the dashboard. The other is the fuel gauge. Cash ROAS is the audit that keeps the dashboard honest.

Why do the two numbers disagree?

Five reasons, and most accounts have all five at once.

Windows and definitions keep moving

Attribution is a rule about which sales an ad gets credit for, and Meta has been changing the rules. From January 12, 2026 the 7-day view and 28-day view windows stopped returning data in the Ads Insights API, leaving 1-day click, 7-day click, 28-day click, 1-day engaged view and 1-day view.⁴ Meta's stated aim was to reduce discrepancies with Ads Manager.¹

Then on March 3, 2026 Meta said click-through attribution for website and in-store conversions would count link clicks only, with shares, saves and other non-link clicks moving to engage-through attribution, and the video engaged-view threshold dropping from 10 seconds to 5.⁵ If you compare this quarter's platform ROAS with last year's, you are comparing two different definitions. I walked through the earlier change in View-Through Is Gone: Reading Meta Results After January 12.

Credit is not causation

An ad can be credited for a sale that would have happened anyway. Meta's incremental attribution setting tries to estimate the true lift. In a test of more than six accounts and about $1.05 million in spend, Seer Interactive found Meta reporting that 87% of April 2025 conversions were incremental, while GA4 showed 67% for the same accounts.²

That is one agency's portfolio, so it may not match yours. The point is the 20-point gap between two tools looking at the same sales. I covered how to test this properly in Incremental Attribution: Measuring What Meta Actually Caused.

Barry Hott, who has run Meta ads since 2008 and sells a course on the topic, argued the same in a video he posted on May 2, 2025.⁶

Plays here, from YouTube

I would go one step further. Do not chase the match at all. Chase a stable ratio between the two.

Refunds, chargebacks and failed payments

A purchase event fires when a card is submitted. Cash arrives when the payment clears and stays. Refunds, disputes and declined renewals live in the gap, and Meta never hears about them. For subscription and telehealth offers the gap can be large, which is why I keep telehealth chargebacks on the same weekly sheet as spend.

Timing

Booked revenue is not collected revenue. A payment plan, a financed sale or an annual contract shows as a full value in Ads Manager and as a trickle in the bank. Read ROAS by cohort, 7 days, 30 days and 90 days after the click, or you will judge a campaign before its cash has landed.

The same sale, claimed twice

Email, search, a retargeting ad and Meta can each claim one order. Add up every channel's reported revenue and you will often exceed your real revenue. Your bank only counts it once.

Which number should set your budget?

Cash ROAS sets the ceiling. Platform ROAS picks the winners underneath it.

Here is a deliberately simple example. These figures are illustrative arithmetic, not data from any account. Say you spend $10,000 in a month and Ads Manager shows $30,000 in purchase value, a platform ROAS of 3.0. After refunds, failed payments and sales that your CRM shows came from other sources, the bank shows $21,000 from ad-sourced customers, a cash ROAS of 2.1. If your gross margin is 60%, the platform view suggests $18,000 of gross profit on $10,000 spent. The cash view says $12,600. Same month, same ads, a very different decision about whether to scale.

Charley Tichenor frames ad tests with the same business-first logic in a video he posted on January 11, 2026. He also sells ad courses, so take the structure as his own, but the test is hard to argue with: "The only question we need to ask is did total revenue minus ad [spend] get better?"⁷ That is cash thinking applied to a test. Judge the change by what the whole business kept.

Meta's own results hint at why the denominator deserves attention too. In Q1 2026 ad impressions grew 19% year over year and the average price per ad rose 12%.³ Those are platform-wide figures, not your account, but they are a reminder that the cost side of ROAS moves independently of how good your ads are.

How do you reconcile them each week?

With a one-hour routine you actually repeat. The aim is a stable ratio you trust, not a perfect match.

I build this into the reporting layer of the system I build for clients, because a ratio you check every week catches an attribution change in days instead of at the end of a quarter. The ratio is what matters. If it jumps after a Meta definition change, you will know the cause is the ruler and not the business.

What if you cannot tag every sale?

Do it by cohort instead. Compare total new-customer cash against total ad spend for the month, then compare to the platform's number. It is a blunter instrument, but it still catches a drift. Add a "how did you hear about us" question at checkout to fill in what attribution cannot see.

What should you do first this week?

Start with the smallest version, and write the two numbers side by side.

Pull last month's platform ROAS and last month's cash ROAS and put them in one cell each. Note the gap. Then ask which of the five causes explains most of it in your business. Subscription offers usually find refunds and timing. Retargeting-heavy accounts usually find credit without causation.

Once the weekly sheet is steady, you also know how much cash really clears, and that is the number a long-term plan is built on, not the one in Ads Manager.

You do not need to distrust Ads Manager. You need to stop asking it for something it cannot measure. It knows what it showed and who clicked. Your bank knows what was paid. Run the business on the number that clears.

Keep building.Ekai