The Conversions API
What sending data back actually does to the algorithm.
The first CAPI setup I ever audited had a match quality score of 3.1. The agency had "installed CAPI" six months earlier and billed for it. It was forwarding the pixel's own broken events through a server, with no email, no click ID and no order ID. Meta was receiving twice the purchases and matching half of them.
The one-sentence version
The Conversions API is a way to tell Meta "this person bought" from your server, with enough detail that Meta can work out who. The transport is easy. The detail is the whole game.
The setup
Build the message
A purchase event is a small package: what happened, when, and a set of match keys describing who. Meta scores the package from 0 to 10 on how well it can connect it to a real account. Below 6, most events are wasted.
Try it: build the event Meta receives
What goes in the message decides what Meta can use
Event match quality
2.5 / 10
Purchases Meta can attribute
50 of 100
No order ID means Meta cannot de-duplicate this event against the pixel. Every purchase is now counted twice. This is the most common CAPI mistake.
CAPI is not a checkbox. It is what you put in it.
Why it matters
Better events, cheaper customers
Meta's delivery learns from the purchases it can attribute. Feed it 40 well-matched purchases instead of 20 fuzzy ones and it has twice the examples of who buys. That is the mechanism behind "sendback improves performance": not magic, just more and cleaner training data.
What good sendback needs
Server-side capture (so the event exists), identity at checkout (so there are keys), order ID (so it de-duplicates), and verified attribution (so you do not send Meta a purchase Google actually drove).
What bad sendback does
Doubles your purchase counts, inflates ROAS, teaches Meta to chase existing customers, and makes the next audit painful.
The agency's 3.1 became 8.4 in a week once we sent hashed email, the click ID and the order ID. Cost per purchase moved three weeks later. Same ads.
The TrueROAS version
Verified in, verified out
TrueROAS only sends purchases that exist as Shopify orders, with the full key set, de-duplicated by order ID, to Meta's Conversions API and Google's enhanced conversions. If attribution says Google drove the order, Meta does not get told it was theirs. That last part is what most CAPI tools skip.
Honest footnote
When CAPI will not save you
If your campaigns are under-delivering because of creative, audience or offer, better events will not fix that. CAPI fixes signal. It does not fix ads. Check your verified ROAS per creative before you blame the pipes.
Questions people actually ask
Straight answers
Does Shopify's native Meta integration already do this?
It sends purchase events, often with limited match keys and without verified attribution. Run TrueROAS alongside it; events are de-duplicated by event ID so nothing doubles.
Is there a Google equivalent?
Yes: enhanced conversions (hashed customer data on the tag) and offline conversion import (upload conversions with the click ID). TrueROAS uses both.
Will my reported ROAS go up or down?
Both, depending on the campaign. Prospecting usually gains attributed purchases. Retargeting often loses some because duplicates and view-only claims stop counting. Judge on blended MER and new-customer revenue.
How long until it affects performance?
Match quality moves within days. Cost per purchase typically follows over one to three weeks as delivery relearns from better events.
See your own numbers, verified against your orders.
Free under $10k/month in revenue. 14-day trial above that. Nothing changes in your ad accounts. You will know within a week which platform has been telling stories.
"I built this because I was making $50k-a-month decisions on numbers I knew were wrong. If TrueROAS does not show you something Meta or Google got wrong in the first two weeks, tell me and I will help you cancel."
Rasmus Arvidsson
Founder, TrueROAS. Answers support himself.
Keep going