Did your Meta ads actually pay back? (and how to know for sure)
Your Ads Manager says last month's campaign hit 3.2x ROAS. So the spend paid back, right?
Probably. Maybe. Honestly, you can't tell from that number alone.
ROAS in Ads Manager is the platform's own attribution — it counts conversions that happen inside its window, on its terms, with its definition of a conversion. It is a useful signal. It is not the answer to "did the money I spent bring back more money than it cost."
This is the gap most teams run into once they spend enough on Meta to care: the dashboard says the ads worked, finance says revenue is flat, and nobody can reconcile the two. The fix isn't a better ROAS number. It's a different question.
What platform ROAS actually measures
Meta attributes a conversion to your ad if the conversion event fires within the configured attribution window (default 1-day click or 7-day view, depending on your setup and consent mode). The conversion event is whatever you defined — a purchase, a lead form, a "complete registration." Two things follow:
The window is short. A buyer who clicked your ad on Monday and bought on Friday via organic search might land inside a 7-day click window — or might not, depending on whether the click was the last touch and whether consent mode degraded the signal.
The conversion is self-reported. The event fires from your pixel or the Conversions API. If your pixel fires on a "thank you" page that loads twice, you get two conversions. If it fires before payment succeeds, you count a conversion that never became revenue.
None of this is Meta being deceptive. It's the platform doing platform attribution, which is designed to optimize delivery, not to close your books.
The question you actually want answered
Stop asking "what's my ROAS?" and ask three smaller questions:
Did the spend bring in new buyers — or did it mostly re-acquire people who already knew you?
Did the revenue those buyers generated exceed the spend, after COGS, not before?
How long did it take for the revenue to arrive?
The first is a customer-acquisition question. The second is a unit-economics question. The third is a payback-period question. ROAS muddles all three into one number that's easy to report and hard to trust.
What you need to answer them
You need three streams of data joined on a shared identity:
Spend — actual dollars out the door to Meta, by campaign and day.
Funnel events — the real things people did on your site or in your product: sessions, add-to-cart, checkout, the conversion itself, by source.
Revenue — the money that actually landed in your bank or your database, tied to the buyer.
The join key is identity — a gclid or a utm pair captured at first touch, resolved to a person, and carried through to the order. When a buyer who first arrived on a Meta ad places an order two weeks later, that revenue attributes to the spend that acquired them. Not because Meta says so. Because the data says so.
This is what "did the spend pay back" actually means: spend in, revenue out, joined by who the buyer is and where they came from.
A worked example
You spend $4,000 on Meta in a week. Ads Manager reports 47 purchases and 3.2x ROAS — call it $12,800 in attributed revenue. Looks great.
Now join it to your database. Of those 47 purchases, 31 are first-time buyers and 16 are returning customers who would likely have bought anyway. The 31 new buyers generate $8,900 in first-order revenue, and over the next 60 days another $4,200 in repeat revenue. COGS on that $13,100 is 55%, so gross profit is $5,895.
Spend $4,000, gross profit $5,895. The campaign paid back — but at a 1.47x gross-profit multiple, not 3.2x ROAS, and it took 60 days to get there. That's the honest number. It's still a win. It's just a smaller, slower win than the dashboard promised, and now you can decide whether to scale it with real numbers.
Where FunnelKeeper fits
FunnelKeeper exists to do exactly this join without you building it. It pulls Meta (and Google) spend, ingests GA4 traffic and funnel events, and reads conversions and revenue from your database — then answers "did the spend pay back" per product, per channel, per cohort. Unattributed revenue shows up as its own honest category instead of being smeared across channels to make the numbers look tidy.
You can see how it works or try it on your own data.
FAQ
So is ROAS useless? No. It's a delivery-optimization signal and a rough health check. It's just not a payback number. Use it to decide which campaigns to keep feeding; use joined spend-to-revenue to decide whether the channel is worth scaling.
What if I can't track identity to the order? You can still get most of the way with a gclid/UTM captured at first touch and a conversion event that carries it. Full identity-to-revenue needs your database, which is why FunnelKeeper reads from it directly rather than relying on the pixel alone.
Doesn't Meta's attribution window overcount? Sometimes, and undercount too — consent mode and SKAdNetwork cut both ways. The point of joining to your own revenue is that you stop depending on the platform's window at all.