How platform ROAS is calculated
Every ad platform divides the revenue it believes it caused by the money you spent inside it. Meta Ads Manager does this with data from its own pixel and Conversions API, its own attribution window, and its own rule about whether a view counts as well as a click. Google Ads does the same with different rules. The arithmetic is trivial; the whole argument sits in the word caused.
Two settings do most of the work. The attribution window decides how long after seeing or clicking an ad a purchase still counts. The click-or-view setting decides whether someone who merely scrolled past the ad and bought later is credited to it. Widen either and the reported figure rises without a single extra sale being made.
Why platform ROAS matters
It is still the fastest signal you have inside the account. Comparing two ad sets that share the same pixel, the same window and the same catalogue is a fair comparison, because roughly the same measurement error applies to both sides. That is enough to decide which creative to keep and which audience to drop, which is what most day-to-day optimisation actually consists of.
The trouble starts when the number leaves the account. Once platform ROAS appears in a monthly report beside real revenue, it is being asked to do a job it cannot do: state what the business earned.
Where platform ROAS goes wrong
Retargeting inflates it most. An audience that already visited your site was going to come back for some of those purchases anyway, so credit for sales you would have made regardless lands in a campaign that then looks outstanding. Running Meta and Google together makes it worse, because both can claim the same order; adding their reported revenue together produces a total the bank never saw.
Returns, cancellations and failed payments are the quiet part. The platform records the purchase event and never hears about the refund. Where cash on delivery is common, as it is across much of Nepal, a meaningful share of recorded orders is never collected at all, and platform ROAS carries every one of them as revenue.
How to act on it
Keep platform ROAS for decisions within the platform, and judge the business on money you can bank. The marketing efficiency ratio — total revenue divided by total advertising spend — cannot double-count, because there is one revenue figure and one spend figure. Read it monthly against what your shop or CRM actually settled, not against what the dashboards claimed.
When a channel looks too good to be true, test it rather than argue about it. An incrementality test that pauses the campaign for a matched group tells you what the platform’s own reporting never can. Decide how your Meta ads will be measured before the budget is committed, not after the first flattering report arrives.