1. Using revenue with tax included
Always feed the calculator net revenue, excluding VAT or sales tax — that money was never yours. At a 21% rate, using gross revenue inflates your ROAS by about a fifth, which is more than enough to keep a losing campaign running for a month. If your store reports gross, divide by 1 + the rate before you type it in.
2. Forgetting returns and refunds
A 3× ROAS with a 20% return rate is really a 2.4× ROAS. Platforms report the purchase, not the refund that arrives three weeks later, so the number you see is always the optimistic version. Apply your historical return rate to the revenue figure before entering it, and remember the returned goods often cannot be resold at full price.
3. Ignoring payment and platform fees
Add them to COGS. Two or three percent sounds trivial until you notice it is a tenth of a 25% margin. Payment processing, marketplace commissions, currency conversion and chargeback fees all scale with orders, which makes them variable costs, which puts them squarely in COGS.
4. Trusting platform-attributed revenue blindly
Meta and TikTok will both claim the same sale, and Google will claim it a third time. Add up the revenue every platform reports and it routinely exceeds what your store actually took. Cross-check the total against real store revenue for the same period, and scale each platform's figure down proportionally before judging any single campaign.
Attribution windows matter as much as double-counting. A 7-day-click, 1-day-view window tells a very different story from 1-day-click, and neither is wrong — but comparing a campaign measured one way against a campaign measured the other is.
5. Judging a campaign too early
A ROAS measured after two days on a 30-day consideration cycle is noise, not data. Give a campaign enough conversions to be statistically meaningful — as a rough floor, 50 conversions before you draw conclusions, and a full purchase cycle before you declare it dead. Killing campaigns on day two is the most common way to burn a testing budget without learning anything.
A sixth, quieter one: comparing different levels
Taking spend from one ad set and revenue from the whole campaign produces a number that looks fantastic and means nothing. Always pull both figures from the same level and the same date range. It sounds obvious; it happens constantly in month-end reporting.
The habit that fixes most of this
Once a month, reconcile the platform's numbers against your accounts: real revenue net of tax and refunds, real cost of goods, real ad spend including agency fees. Use those figures in the calculator to get a true ROI, then work out what ROAS the platform needs to report for that true ROI to hold. That is the target you hand your media buyer — and it is usually higher than the one they have been working to.