How purchase ROAS is calculated
Meta adds up the value attached to every purchase event it attributes to an ad, then divides that total by what the ad set spent. The output is a return per unit of currency spent. Everything rests on two inputs: whether your site sends an accurate value with each purchase, and which attribution setting is in force when the totals are drawn.
The value itself is yours to define. Whether it includes shipping, tax or discounts depends entirely on what your website chooses to send, and Meta has no way to check. Two shops trading identically can report very different returns purely because one sends order totals and the other sends net product value.
Why purchase ROAS matters
It is the fastest read on whether an ecommerce campaign is paying for itself, and it is the figure that value optimisation aims at. When you set a return target, delivery starts favouring people it expects to spend more rather than people it merely expects to buy, which changes who sees the ads and what they are shown.
It is also a useful early warning. A return sliding while order volume holds usually means the mix has shifted towards cheaper items or discounted stock, which is a merchandising signal as much as an advertising one.
Where purchase ROAS goes wrong
The first problem is that it is revenue, not profit. A return that looks comfortable against spend can still be a loss once cost of goods, shipping, payment fees and returns come out. Every business has a break-even return, and it is a different figure for every margin.
The second is credit that is not exclusive. Meta reports what it can attribute inside its own window. Google Ads does the same inside its own. A single customer who met both appears in both reports, so adding the platform figures together describes more revenue than the shop actually took.
The third is incrementality. Returning customers and people already searching for the brand carry high value and would often have bought regardless. Retargeting campaigns look strongest for exactly this reason, and the reported return overstates what that spend genuinely added.
How to act on it
Anchor the platform figure to reality. Compare total revenue from your shop’s own reports against total advertising spend for the same period; that blended view cannot be double-counted and shows whether the business is actually ahead. Use the platform figure to compare campaigns against each other, and the blended figure to decide how much to spend overall.
Set the target from margin rather than ambition. Work out the return at which a sale stops being worth making once goods, delivery, fees and returns are paid, then keep the campaign target above that with room to spare. And check what value your site is sending before trusting any of it, because a purchase event with a missing or wrong value quietly ruins every calculation built on top of it.