Ground truth
The platform can't see the number that decides it.
An ad platform optimizes toward what its own pixel can measure. It never sees your cost of goods, your return rate, or what a customer is worth on the second order. So when it finds a campaign that converts cheaply and loses money on every order, it does the logical thing: it scales it. Nobody is asleep at the wheel — the dashboard is green, the target is met, and the report is correct on its own terms. That is the expensive part.
Conversions are up and return on ad spend is above target.
The volume is discount-driven. Thin contribution per order, and these buyers rarely come back.
Same campaign. Two verdicts. Only one of them is looking at your profit — and it isn't the one spending your budget.
It is your best campaign at acquiring volume and your worst at acquiring profit. The platform can only see the first half.
The disagreement is the finding, not an error.
It is a structural gap, not a feature gap.
This is worth being precise about, because it is the difference between a thing that gets patched in the next release and a thing that cannot be patched at all.
Your cost of goods lives in your store or your ERP. Your return rate shows up weeks after the click. Your second-order value lives in a cohort nobody has sent to the ad platform, and mostly should not.
An optimizer grading its own spend is being asked to mark its own homework. Even a perfectly honest one is measuring the objective it was given, and that objective is volume of a measurable event — not profit.
This is what makes it expensive. Nothing looks broken. There is no error state, no alert, no red number — just a well-run campaign steadily scaling something that loses money per order.
Which is why waiting a quarter is not neutral. Nothing in the account will surface this on its own — the 3.2× campaign above keeps clearing its target, keeps earning budget, and keeps teaching the optimizer that discount-driven volume is what you want more of. Ninety days of that is not ninety days of no decision. It is ninety days of one decision, made by the system with the least visibility into your profit.
The platform grades its own homework, and it passes.
Every number it reports is true. None of them is the one that decides whether a campaign is worth running.
- 3.2×the ROAS the platform reports on a campaign losing contribution
- 41%average discount on the orders behind it
- 8%of those orders came back — neither figure reaches the bidding
Reconciled against the numbers your business runs on.
Adgent reads GA4 and Search Console alongside the ad platforms, so a report is checked against a source the platform does not control — and where the two disagree, the disagreement itself is the finding. Shopify reading is the next connection we are building.
Reads are read-only. A connector is listed here only once it is running in the product — the ones marked next are being built, and we would rather show you the line than blur it.
Store revenue, not platform-reported revenue
Platform-attributed conversions and what your store recorded are two different numbers, and the gap between them is usually the whole argument. Both are shown, with the gap named.
Attribution normalized across platforms
Meta and Google count a conversion over different windows with different rules. Comparing them raw produces a clean-looking chart that is not a comparison. Adgent puts them into one comparable shape first.
The verdict follows the money
Once the numbers reconcile, the position changes. "Scale this" and "stop this" can be the correct call on the same campaign depending on which lens you are looking through — and only one of those lenses is looking at your profit.
Discount-driven volume is the classic case: cheap conversions, thin contribution per order, buyers who rarely come back. Every measurable signal says winner. The only lens that disagrees is the one the platform does not have.
Ground truth is what makes a verdict worth reading.
Reconciled numbers are the input; a position you can act on is the output. And the reason no in-platform tool closes this gap is the same reason the comparison page exists.
Find out how far your platform numbers are from your numbers.
Give us read access and we’ll put platform-reported revenue next to what your analytics recorded for the same window, and name the gap.
- Platform figures against GA4 and Search Console
- Attribution normalized across Meta and Google first
- Where they disagree, the disagreement is the finding
The size of the gap, campaign by campaign, and which of your winners change places once the numbers reconcile.
Thanks — we’ve got it.
We’ll be in touch within a couple of working days.