Incremental.ai measures what your marketing actually caused
Know which channels actually grew your revenue.
Meta says it drove the sale. Google takes credit too. GA4 disagrees, and Shopify reports something else entirely.
Once a month you get one report that settles it, so the next dollar goes where it actually earns.
Who claims what, in one example month
You are budgeting against revenue that was never there
Add up what your channels claim and the total is larger than what your store took. Every one of them is honest about its own corner, and the overlap is invisible from inside any of them.
That gap is not an accounting curiosity. It is the number your budget is being allocated on.
Why you can act on it
A number you can defend to your board.
Seasonality, the promo, demand that was already climbing. A channel only gets credit for a sale once the ordinary explanations have failed to account for it, and the reasoning is on the page rather than behind a model nobody can question.
- 01
Put every source on one basis
Platforms count differently. Line them up first, so what still disagrees is a real disagreement.
The reasoning
Platforms count differently: different windows, different rules for what counts as a sale, different time zones. Every source is normalised and reconciled against your order ledger first. What disagrees after that is a real disagreement, not an accounting artifact.
- 02
Treat each source as a witness
A platform reporting on its own ads has a stake in the answer. Its testimony gets weighed, not accepted.
The reasoning
A witness can be honest and still be wrong, and a witness with a stake remembers things in its own favour. A platform reporting on its own ads has that stake, so its testimony is weighed rather than accepted.
- 03
Collapse the same story told twice
One event moving down one path is one witness, not four. More data of the same kind cannot raise confidence.
The reasoning
When the ad fires, the click follows, the session follows, the order follows. That looks like several tools agreeing. It is one event moving down one path. Signals sharing a source count as roughly one witness, so more data of the same kind cannot raise confidence.
- 04
Argue the boring explanations first
Seasonality, your promo, demand that was already climbing. Cleared every month, not when someone remembers.
The reasoning
Seasonality, the promo you ran, a stockout clearing, demand that was already climbing before you spent a dollar. These are eliminated by construction every month, not when an analyst remembers to check. Only what survives them competes to be the answer.
- 05
Get suspicious when it is too tidy
Perfect agreement is what a shared cause looks like. So it widens the range instead of narrowing it.
The reasoning
If a shared cause moved underneath the month, every signal rises together whether or not your ads did anything. Perfect agreement is what a common cause produces. So unusually clean agreement widens the range instead of narrowing it.
- 06
Conclude once, then earn the confidence
The number is settled first. How far to trust it is decided afterwards, and cannot reach back to flatter it.
The reasoning
The estimate is produced in one place. How much to trust it is decided separately, afterwards, and cannot feed back into the number to make it look better. Without an ingested experiment the read stays associational and capped, and the report says so.
A monthly answer, and a daily view for the team
The report lands on the 3rd of every month. The same connections that produce it also fill a workspace your team can open any morning: contribution margin in dollars, the goals you set, and every creative judged against your own account rather than ranked and handed back.
The workspace reports what each source said and claims nothing about cause. Working out what actually caused what is the report's job, and the report is the thing you are buying. There is no separate plan and no add-on: connect your store and your ad accounts and you get both.
Net sales
$214,000
+12.4%vs the previous 28 days
Pace $236,000
MER
3.40×
+0.31vs the previous 28 days
Target 3.10× · 110%
Contribution margin
$87,740
+9.1%vs the previous 28 days
41.0% of net sales
CAC
$48.20
−4.8%vs the previous 28 days
Target $52.00 · 108%
Read a real report.
The year so far, every channel, exactly as a paying brand gets it. No email, no call.