Why Your Ad Reports Show 3 Different Numbers: Meta vs Google vs Shopify

This piece is part of the Honest Audit Manifesto, the full BTB worldview on what is wrong with the DTC ad audit category.

Open your Meta Ads Manager. Note the conversions for the last 30 days. Now open Google Ads. Note the same window. Now open Shopify. Note the orders. Three different numbers. They do not match. The gap is often 20 to 40 percent.

Every operator at $20K+ in monthly spend has seen this moment. Most are told it is "an attribution problem" by their agency or by an attribution SaaS (Software as a Service) pitching a fourth number you can trust. The honest answer is that the gap is structural and intentional. Each platform reports the version of the truth that justifies the most spend on its own properties. Understanding the mechanics is the only way to make sense of any agency report you will ever read.

Meta's own documentation makes the over-attribution explicit. Meta's attribution system help page confirms the default model credits a conversion when the user clicked an ad within 7 days or only viewed an ad within 1 day. That single design choice is the source of the inflation. A user can see one Meta ad, never click, buy through Google a day later, and Meta still claims the conversion. The math is built to flatter Meta.

The 3 platforms, the 3 attribution models

Each platform's attribution model is a design decision, not a measurement standard. Each design supports a specific commercial outcome. Walk through them in order and the gap stops being a mystery.

How Meta measures conversions

Meta's default is 7-day-click plus 1-day-view. In plain language:

• If a user clicks a Meta ad and buys anywhere within 7 days, Meta claims the conversion.
• If a user only views a Meta ad (no click) and buys within 24 hours, Meta claims the conversion.
• The two windows overlap, so a single user can be counted across multiple campaigns inside the same account.

Why Meta uses this model: it captures the maximum credit Meta can defend. Wider windows mean more claimed conversions. More claimed conversions mean more justified budget. View-through credit is the most aggressive part. A user who saw an ad once, did not click, and bought through a different channel still gets logged as a Meta conversion.

For the upstream mechanics of how your Meta Pixel and Conversions API feed into this attribution system, see Stage 2 connections and event firing in the Meta audit method. The technical layer underneath this attribution model is the Pixel + CAPI deduplication. For the operator-level setup guide, see the Conversions API setup guide.

How Google measures conversions

Google's default since 2024 is data-driven attribution across Google properties. In plain language:

• Google credits conversions across the Google journey (Search, YouTube, Display, Gmail, Maps).
• Google only sees touchpoints inside Google's own properties.
• A user who saw a Meta ad on Monday and a Google ad on Friday is credited by Google for Friday. Meta is invisible to Google's model.
• Google does not deduplicate against Meta's claim.

Google's data-driven attribution documentation confirms the model "looks at all the interactions, including clicks and video engagements, on your Search, YouTube, Display, and Demand Gen ads in Google Ads." That phrasing is doing real work. Everything outside Google is excluded by design.

Why Google uses this model: it captures Google's full influence inside its own walled garden. Acknowledging external touchpoints would shrink Google's reported impact and weaken the case for Google spend. For how this plays out at the campaign level, see the Google Ads audit method.

How Shopify measures conversions

Shopify reports order-completion. In plain language:

• Shopify counts actual transactions that closed in the store.
• Shopify's "Sales attributed to marketing" view uses Shopify's own attribution logic, which differs from both Meta and Google.
• Shopify cannot see Meta or Google ad impressions. It only sees the click that arrived on the store, plus the order data.
• Shopify shifts between first-click, last-click, and assisted-conversion logic depending on which report you open.

Why Shopify uses this model: it positions Shopify as the merchant's source of truth, which strengthens Shopify's brand against WooCommerce and BigCommerce. The number is closer to reality because it sits next to the money. But Shopify is not neutral either. Its attribution logic still has gaps, especially around view-through influence and longer consideration cycles.

The differences are not edge cases. They are the design.

The math of the gap

Here is the simplest worked example. A brand running $20K Meta plus $10K Google for one month.

Plug in your own last-30-days numbers and see where your account sits on the same curve:

The discrepancy is not noise. It is mathematically guaranteed once you know the attribution windows and models. Meta's window is wide and includes view-through. Google's window is wide but blind to Meta. Shopify is narrow but actual. The three numbers cannot reconcile because the three platforms are looking at different slices of the same purchase, and two of them are looking at slices that overlap.

The same pattern explains why the platform-summed ROAS (return on ad spend) on a typical agency report runs 40 to 70 percent higher than the blended ROAS calculated from Shopify revenue divided by total ad spend. The agency report is summing claimed credit. The blended math is reading actual money.

This is the platform-layer version of the structural problem. For the agency-layer version, where agencies sit comfortable behind the inflated combined number, see how agencies exploit these platform-reporting gaps to inflate their reports.

4 signals your reports are diverging more than they should

Pull these tonight. The work takes 30 minutes. Each signal can be checked from your own dashboards. No agency access required.

Any one of the four signals is enough to question the agency report. Two or more, and the reported ROAS is almost certainly inflated by 30 percent or more.

In one audit on a DTC (direct-to-consumer) skincare brand running $45K monthly across Meta and Google, signals 1 and 4 both fired. Meta plus Google claimed 8,200 conversions. Shopify showed 6,100 orders. That is 34 percent inflation. The agency had been reporting 4.3x ROAS. The blended number from Shopify revenue divided by total spend was 2.7x. The gap was structural, not bad luck.

The honest reframe: stop chasing a single true number

The mistake most operators make is looking for a fourth number to call "the real one." There is no real one. Each platform is a partial view. The work is to know the gaps in each view and reconcile to the bank account.

Here are the 4 moves that actually help.

1. Use Shopify (or your cart) as the revenue source of truth. Whatever Shopify shows is what actually happened. Anything else is a claim about what happened.

2. Calculate blended ROAS. Total ad spend across all platforms, divided by total revenue from Shopify. This number cannot be inflated by attribution overlap. It is the only honest paid-channel number. For the gross-margin math behind what counts as a good blended ROAS, see the gross-margin math behind a good ROAS.

3. Evaluate at the P&L level, not the platform level. What matters is gross profit after ad spend, COGS (cost of goods sold), returns, shipping, and overhead. Not the ROAS any single platform reports. If your P&L says ads are profitable, they are. If it says break-even, the platform reports do not change that.

4. Treat platform reports as inputs, not answers. Meta's number tells you what Meta is claiming. Google's number tells you what Google is claiming. Shopify's number tells you what closed. Use the three together to triangulate. Never trust any one of them in isolation.

The closing position is simple. There is no neutral observer in performance marketing attribution. Meta benefits from inflated credit. Google benefits from defending its walled garden. Shopify benefits from being source-of-truth even when its own data has gaps. The only thing that does not have a commercial incentive in this conversation is your P&L. Treat that as the answer, and everything else as evidence.

Platform inflation is one half of the ROAS math problem. The other half is the LTV-adjusted reality. See why you're calculating your ROAS wrong.

Why an attribution SaaS does not solve this

A whole industry exists to sell you a fourth number. Triple Whale, Northbeam, HockeyStack, Polar Analytics, Wicked Reports, and a dozen others. They all promise to "solve attribution" by stitching cross-platform data into one model. Some of them are useful tools. None of them are neutral.

The SaaS model has its own incentive structure. The vendor wants you to keep paying the monthly fee. The model is built to surface insights that justify the subscription. The methodology is proprietary, which means it is not auditable in the same way Meta or Google or Shopify reports are. When the SaaS-attributed ROAS conflicts with your P&L, you are stuck choosing between two numbers, neither of which lives next to your money.

That does not mean attribution tools are useless. It does mean they are not a replacement for blended ROAS calculated from your cart revenue. Use them for directional insight on creative testing or audience overlap. Do not use them as the single source of truth that justifies your retainer or your budget.

The fourth number is also constructed against an incentive structure. The closest thing to neutral is your own P&L, and even that has assumptions baked in.

The structural critique extends past attribution into the rest of the funnel. The same divergence between reported and actual numbers shows up in checkout. For the page-speed leak that compounds with attribution leaks, see the mobile page speed leak that compounds with attribution leaks.

Keep reading on attribution, the audit method, and the agency layer