Problems we solve / The marketing / Who gets the credit
Problem 09 of 16 · The marketingMeta, Google and Klaviyo all claim the same sale. Who actually gets the credit?
Add up the conversions Meta, Google and Klaviyo report and the total is bigger than the orders Shopify actually took. Every platform grades its own homework with an attribution model that favours itself. Marketing attribution only becomes useful when one answer is reconciled to the orders and the margin that really arrived.
Stop picking a platform to trust. Compare the attribution models side by side, check each claim against the contribution margin that arrived, and publish one source per customer that the whole business reads. Blufire's Attribution section does that with the Model Comparison Matrix, Margin Triangulation and a Canonical Source Map.
5.0 on Google · 100+ businesses · $153M revenue influenced
One month, three dashboards, one Shopify.
A store took 1,000 orders in a month at an average order value of A$90, on A$30,000 of total ad and email spend. Here is what each platform reported as its own.
Add the platforms' own columns and the spend looks like it returned about 4.0x (A$119,700 ÷ A$30,000). The business actually banked 3.0x on revenue (A$90,000 ÷ A$30,000), and less again once margin is counted. The 330 is a floor: some orders came from organic search or direct and were claimed by nobody, so the real double counting is larger.
Budget set from platform dashboards drifts toward whichever platform claims most aggressively. The honest cross-check is blended, total revenue or margin over total spend, which is the MER view. Platform over-claiming walks through the full method.
Every platform counts the sale it touched, not the sale it caused.
A customer sees a Meta video, later clicks a Google search ad, then opens a Klaviyo email and buys. One order. Meta counts it inside its view or click window, Google counts it on the click, Klaviyo counts it on the open. Each dashboard is correct by its own rules, and together they report three sales. The sources cited on The Math put platform claims at up to 140% of actual revenue.
Each platform also models conversions it cannot observe. The same page cites Meta over-reporting by roughly 26% against third-party analytics on average, and Google Ads running 15% to 20% high under modelled conversions (Measured, 2024; Varos, 2024). None of them can see the others, or your costs.
Choosing one attribution model does not fix it. Last click hands the sale to branded search and email, which mostly catch existing demand: The Math cites branded search as 60% to 80% non-incremental. First click hands it to prospecting. Every model is a rule for sharing credit, not a measure of cause, so it has to be checked against the orders and margin that landed, and in the end against incrementality.
One published answer, checked against margin.
Section S5, Attribution, runs the models side by side in the Model Comparison Matrix instead of trusting one. Where they disagree about a channel, you see it, and you see by how much. The Journeys view shows the real touches behind each order, with each step's share of credit under the model you choose.
Margin Triangulation then checks each model's claims against the contribution margin that actually arrived. What gets published is one Canonical Source Map: a frozen first-purchase source per customer that every other section reads, so channel reports, cohort payback and creative scores all argue from the same answer instead of three.
- Model Comparison MatrixAttribution models reconciled side by side, so disagreement between them is visible rather than hidden inside one dashboard.
- Margin TriangulationEach model's claims checked against the margin that actually arrived, not the revenue a platform reports.
- Canonical Source MapOne frozen first-purchase source per customer, published once and read by every other section.
- Journeys and arrival qualityThe real paths customers take, a grade on each channel's arrival quality with drift caught early, and UTM hygiene cleaned up.
Real product screen, shown on sample data.
The team behind the numbers.
A$942kin incremental revenue once the double-counted attribution was fixedRead the case study →“I couldn't be more impressed with the Blufire team and the improvements they have made… working on the account and maximising results daily.”
Getting to one marketing attribution answer.
- Reconcile to Shopify first.Before comparing channels, check the sum of platform claims against real orders. The gap is the size of your over-claiming problem.
- Run more than one model.If a channel only looks good under one model, that is information. Channels that hold up across models deserve more trust than those that do not.
- Split branded from non-branded.Branded search and retargeting catch demand that already exists. Judge them apart from prospecting, and read the three CACs before scaling either.
- Test what the models cannot settle.Where models disagree and the money is large, a geo-lift test or holdout settles it. See did it cause margin.
The decision you walk away with.
Meta, Google and Klaviyo each report their own total, and together they add up to more than Shopify.
One reconciled view in which every order is counted once.
The attribution model is whatever each platform defaults to.
Models compared side by side, with the differences in the open.
Budget follows the platform that claims the most.
Budget follows the margin that actually arrived, channel by channel.
Each report uses a different source for the same customer.
One Canonical Source Map that every section reads.