Meta says 5× ROAS. Here’s how to check it against your real orders

By Gokul, founder of CyberShapers · · 6 min read
A business owner checking paper receipts at a desk

The short answer ROAS means how many rupees of sales came back for each rupee you spent on ads. The number in Meta is Meta’s own estimate, not money in your bank. To check it, compare it with your real orders for the same dates. Then use the calculator below to see whether that return actually makes you a profit.

Think of it like thisImagine paying a salesperson who also writes their own report. They take credit for every customer who walked past their poster, including people who were coming in anyway. The report isn’t a lie, but you’d still check the till.

What the number means

It’s simple division: sales Meta credits to your ads, divided by what you spent. Here’s a real example from an e-commerce account we manage.

How a Meta-reported ROAS of 5.18 is calculated Purchase value of 4,58,790 rupees divided by ad spend of 88,501 rupees gives a 5.18 times ROAS. The same account shows 63 purchases, about 1,405 rupees cost per purchase and about 7,282 rupees average purchase value. All figures are as reported by Meta Ads Manager for 1 to 13 July 2026. ₹4,58,790 sales Meta credits ÷ ₹88,501 ad spend = 5.18× 63purchases ₹1,405per purchase ₹7,282average value Meta-reported, not audited. Last two figures are our maths.
Meta Ads Manager, 1 to 13 July 2026, using Meta’s own rules. One 13-day window, not a guaranteed result.
Meta Ads Manager table showing campaigns, spend and purchase return
Where those numbers come from: the real Ads Manager view. Client name withheld and campaign names shortened.

These numbers haven’t been checked against the client’s orders, and our case study says so. A number like this is where your questions should start.

Why Meta’s number won’t match your orders

What sits between Meta's reported sales and sales your ads really added An illustration with no real data. A full bar shows sales Meta reports. Three slices step down from it: cancelled, refunded or undelivered orders; people who only viewed an ad or would have bought anyway; and tracking differences. A shorter green bar shows the sales your ads really added. Sales Meta reports Cancelled, refunded, COD not delivered Viewed only, or wouldhave bought anyway Tracking differences Sales your ads really added
An illustration, not client data. The size of each slice is different for every business, which is why you check yours.

Meta takes credit for a while afterwards

If someone clicks your ad and buys within days, Meta counts it, even if they came back through Google. A common setting is 7 days after a click, or 1 day after just seeing the ad. Your account shows which setting it uses.

Some orders don’t stay sold

Meta counts the order when it’s placed. It doesn’t automatically learn about a later cancellation, refund or return, or a cash-on-delivery parcel refused at the door. If you sell on COD, check this first.

Some people were buying anyway

Your regular customers see your ads too. If one sees an ad and buys as planned, Meta can still count it. The report looks good, but the ad didn’t add much.

Tracking isn’t perfect

A sale can be counted twice if the tracking is set up badly, or missed if a browser blocks it. Dates can differ too, so a sale can land in a different week in each report.

How to check it in 30 minutes

  1. Pick a month that’s finished, so most returns and cancellations have come through.
  2. Download from Ads Manager: spend, purchases and purchase value.
  3. Download from your store: orders and sales for the same dates. Remove anything cancelled, refunded or undelivered.
  4. Compare the order counts. If Meta claims 60 purchases and you only had 70 orders from all sources, ask how likely that really is.
  5. Work out your real return: total real sales ÷ total ad spend. No one’s estimate is involved. Marketers call this blended ROAS.

A simple testWhen you raise your ad budget, do your real sales go up too? If you double spend and sales barely move, the good ROAS wasn’t real growth.

Does your return actually make a profit?

A 3× return can be great for one business and a loss for another. It depends on what you keep from each sale after product, packing, shipping and payment fees. That’s your margin. Move the sliders to see yours.

Try it

Break-even calculator

You make ₹20 on every ₹100 of ads₹100 of ads brings ₹300 in sales, and you keep ₹120 of that. You break even at 2.5×.

The rule behind it: break-even return = 1 ÷ your margin. Keep 40% of each sale, and you need 2.5×. Keep 25%, and you need 4×.

You keepYou need at least
25%4×
40%2.5×
50%2×
70%1.43×

What to do if the gap is big

  • Check for double counting. Meta’s Events Manager shows whether each purchase is being sent twice.
  • Split new customers from regulars in your store data. Ads that only reach regulars are worth less than they look.
  • Cut COD losses, for example by confirming orders before you dispatch them.
  • Keep using Meta’s numbers to compare ads with each other. Use your own numbers to set your total budget.

Where a dashboard helps

Doing this once is easy. Doing it every week is where it gets dropped. We bring ad spend and sales into one place, so your real return sits next to Meta’s number automatically. See the 12 numbers owners check and Your Business Dashboard in 10 Days.

Questions owners ask

Why does Meta show more sales than my store?

Usually a sale counted twice, an order later cancelled, or a sale dated differently in each report. Check for double counting first.

What’s a good ROAS?

There’s no single number. A good one sits comfortably above your own break-even, month after month. Use the calculator above to find yours.

Should I stop trusting Ads Manager?

No. It’s good for comparing one ad with another. Just don’t treat its sales figure as your income. Your store and bank are the truth for that.

Want a second pair of eyes? Send us a month of Ads Manager and store exports, and we’ll walk through them on a free call.

Request a free call

Keep reading

Request a free call