.svg)
Why Every Ad Platform Claims the Same Sale: The Ecommerce Attribution Problem
Add up the revenue your ad platforms report and you'll often get a number bigger than your actual sales. Meta says it drove the sale. Google says it drove the same sale. GA4 tells a third story. Everyone's ROAS looks great, and yet the bank balance doesn't agree. If you've ever tried to reconcile platform numbers for a scaled ecommerce brand and given up, this article is about why that happens — and what to do instead of trusting any one of them.
The short version: Ecommerce ad platforms overstate their results because each one — Meta, Google, TikTok, your own GA4 — independently claims credit for the same conversions. Each counts a sale if it touched the customer at all (via last-click or view-through attribution), so when you sum every platform's reported revenue, the total exceeds your actual revenue, often by a wide margin. Last-click attribution compounds it by over-crediting the final touch (branded search, retargeting) and starving the top of the funnel that created the demand. The fix isn't a better single number — it's triangulating platform attribution with incrementality testing and marketing mix modeling, so you can see what your ads actually caused.
Why the platforms all claim the same sale
The root cause is simple once you see it: every ad platform is grading its own homework. Meta, Google, TikTok, and the rest each run their own attribution inside their own walled garden, and each is incentivized to take as much credit as possible. If a customer saw a Meta ad and clicked a Google ad before buying, both platforms count the conversion — Meta via view-through attribution, Google via last-click. Neither knows (or cares) about the other.
The result is double- and triple-counting. Sum the revenue each platform reports and it exceeds what you actually made — sometimes dramatically. There's no single ledger; there are several self-interested ones, and "our reported revenue is higher than our real revenue" is the inevitable outcome. It's not a bug you can configure away in any one platform, because the problem lives between them.
The last-click distortion
Layered on top of double-counting is the model most ecommerce brands still default to: last-click attribution, which hands 100% of the credit to the final touch before purchase. That systematically:
- Over-credits the bottom of the funnel — branded search and retargeting look like heroes, when often they're just intercepting demand that other channels already created. Someone who was going to buy anyway searches your brand name and clicks; last-click calls that a "win."
- Starves the top of the funnel — the awareness and prospecting channels that actually generated the demand get little credit, so they look inefficient and get cut. You end up defunding the very channels feeding your "winners."
- Inflates "direct." Traffic that can't be cleanly attributed piles into "direct," which balloons with misattributed sessions and hides where demand really came from.
So even within a single tool, the story is distorted. Combine that with cross-platform double-counting and you have numbers that feel precise and are quietly wrong.
Why this is expensive, not just annoying
This isn't a reporting inconvenience — it's a budget problem. When your attribution over-credits the easy-to-measure channels, you reallocate spend toward them, pulling money from the channels doing the real work. You optimize toward a distorted map, and the misallocation compounds every cycle.
It's part of a broader data-trust crisis. Adverity's 2025 research found that CMOs estimate 45% of the data used to drive marketing decisions is incomplete, inaccurate, or outdated — and not a single CMO rated their data more than 75% reliable. When nearly half your decision fuel is contaminated, attribution is one of the biggest contaminants. And the moment your CFO stops taking platform ROAS at face value — which, increasingly, they do — a last-click dashboard can't prove the difference between revenue you caused and revenue you merely touched.
The fix: stop looking for one true number
The instinct is to hunt for the "right" attribution model or the one tool that finally tells the truth. There isn't one. The modern answer is triangulation — using three complementary methods together, each covering the others' blind spots:
- Platform attribution stays useful as a fast, tactical signal for optimizing creatives and audiences week to week — just never as the arbiter of cross-channel budget.
- Incrementality testing (like geo holdouts) proves what your marketing actually caused versus what would have happened anyway. It's the causal anchor that exposes the channels merely harvesting existing demand. (See our guide to incrementality and how to measure it.)
- Marketing mix modeling (MMM) estimates each channel's real contribution at a portfolio level, using aggregate data that doesn't depend on the cross-platform tracking that's breaking down.
Underneath all three sits the unglamorous prerequisite: a single, owned definition of revenue and clean tracking you'd trust enough to actually cut spend on. Without that foundation, you're triangulating on sand. (For the attribution layer specifically, our guides to tracking marketing channels with UTMs and marketing attribution in Amplitude are good starting points.)
Attribution is one leak of three
Attribution you can't defend is one of three places scaled ecommerce brands quietly lose revenue — the other two being a data foundation no one owns and revenue trapped in the funnel. They compound: unreliable attribution misallocates spend, a shaky data foundation makes every number suspect, and slow testing leaves recoverable revenue on the table.
If the "our platforms report more than we actually made" problem sounds familiar, it's worth seeing where your revenue is leaking across all three. Adasight's free Growth Gap Assessment scores your maturity across data, experimentation, and AI in about three minutes — no email needed to see your result — and shows you the single biggest gap to close first.
See where your revenue is leaking →
Frequently asked questions
Why do Meta and Google report more revenue than I actually made?
Because each platform independently claims credit for conversions it touched, using its own attribution inside its own walled garden. A single sale influenced by both gets counted by both, so summing platform-reported revenue exceeds your real revenue. There's no shared ledger reconciling them.
What's wrong with last-click attribution for ecommerce?
Last-click gives all the credit to the final touch before purchase, which over-credits branded search and retargeting (channels that often just capture existing demand) and starves the top-of-funnel channels that created that demand. It also inflates "direct" with traffic it can't attribute, painting a distorted picture of what's working.
How do I know which channel actually drove a sale?
No single attribution model can tell you reliably. The credible approach is triangulation: use platform attribution for fast tactical signal, incrementality testing to prove causation, and marketing mix modeling for the cross-channel view — then reconcile where they disagree. Incrementality is the method that isolates what a channel truly caused.
What is incrementality testing?
It's a method that withholds marketing from a randomly selected control group (often by geography) and compares outcomes against an exposed group, so the difference is the true causal impact of your marketing. It's the most reliable way to separate revenue you caused from revenue that would have happened anyway.
How much of a problem is unreliable marketing data really?
Significant: Adverity's 2025 research found CMOs estimate about 45% of the data behind their marketing decisions is incomplete, inaccurate, or outdated, and no CMO rated their data more than 75% reliable. Attribution is one of the largest sources of that unreliability, and it directly drives budget misallocation.
