ROAS

Why does the ROAS in the ad platform not match the shop's own?

Because the two count different things. The shop counts every order. The platform counts only the orders it could attribute to an ad it served, so any sale it failed to see is missing from the numerator. The platform's ROAS is therefore usually the lower of the two, and unevenly so.

Uneven is the part that costs money. If one campaign's buyers block tracking more than another's, the two are graded on different amounts of missing data, and the comparison between them stops meaning anything.

Does a higher reported ROAS mean a more profitable campaign?

Not on its own. Reported ROAS moves when attribution improves, even if the campaign did not change and the shop did not sell one unit more. A number that rises because the platform can finally see sales it was already producing is a measurement gain, not a business gain, and the two must not be confused.

What does ROAS leave out?

Cost of goods, shipping, payment fees, returns and the labour behind them. ROAS measures revenue against ad spend and nothing else, so a campaign can hold a healthy ROAS and still lose money on every order. Profit-based measures exist for that reason.