Reconciling Meta, Google and Amazon Ad Data

If you are running Meta, Google and Amazon Ads together and the conversion numbers refuse to reconcile, the reason is structural, not a reporting bug you can fix with a better tool. Meta and Google drive purchases on your own site, which your store or billing system records. Amazon Ads drive purchases that happen on Amazon, on a closed loop your site never observes. Those are two different revenue pools, and no dashboard can add them into one attributed number without inventing data. The setup that works is to stop trying, and to report blended efficiency across the portfolio with each platform measured on its own terms underneath.

This comes up constantly at seven-figure annual spend, where the reconciliation gap is large enough in absolute dollars to stall budget decisions. Below is why the three cannot agree, what each one actually counts, the reporting structure agencies use to present one clear number to a client, and what to do about it this week.

Why Meta, Google and Amazon can never agree

There are three separate problems stacked on top of each other, and most teams only ever diagnose the first one.

Different attribution windows. Each platform decides for itself how long after an ad interaction it will still claim a sale. The windows are not close to each other, and two of the three changed recently.

A single customer who clicks a Google ad on day 1, sees a Meta ad on day 5 and buys on day 10 appears in Google Ads and in Meta, and appears once in Shopify. Add the platform numbers together and you have counted that person twice.

Overlapping credit. Every platform is incentivized to claim as much of the journey as its window allows, and none of them know about each other. This is why the sum of platform-reported revenue routinely exceeds actual recorded revenue by a wide margin. Summing across ad accounts is the single most common error in weekly marketing reports, and it is the one that produces a number no system on earth agrees with.

The Amazon wall. This is the part that makes a Meta plus Google plus Amazon stack genuinely different from a two-platform stack, and it is rarely stated plainly. Amazon purchases complete on Amazon. Your Shopify or Stripe account never sees that order, so there is no shared source of truth to reconcile against. Amazon reports sales back to you in aggregate through ACOS and TACOS, but it does not hand outside tools the customer-level journey, which means no third-party platform can legitimately stitch an Amazon sale into a cross-channel attribution model. Any vendor claiming otherwise is modeling, not measuring, and you should ask which one they mean. We covered the underlying mechanics for the on-site channels in why ad platform numbers do not match.

I am spending $150k a month across three ad platforms and I cannot reconcile the conversion numbers. What tools solve this?

Honestly: none of them solve it as stated, and the vendors who say they do are selling a model rather than a measurement. What tools can do is give you a reporting structure where the irreconcilable parts stop mattering for the decisions you actually make. That structure has three layers.

Layer one, the blended number. Total revenue divided by total media spend, across everything. This is your marketing efficiency ratio, and its virtue is that it is arithmetically incapable of double counting, because there is one numerator and one denominator. At $150k a month in spend this is the number that belongs in the budget meeting. The mechanics, including how to set a break-even MER from your contribution margin, are in our marketing efficiency ratio dashboard.

Layer two, revenue pool separation. Split direct-to-consumer from Amazon and never merge them. Your site revenue reconciles against Meta and Google spend using blended CAC and MER. Amazon revenue reconciles against Amazon spend using ACOS and TACOS, which are Amazon-native and perfectly good at their job. Two clean ratios beat one dirty one.

Layer three, platform ROAS as an optimization signal only. Keep it, label it clearly as platform-reported, and never sum it. It is still the only view with campaign and creative granularity, so media buyers need it daily. It just has no business in a revenue forecast.

On tooling, the category splits three ways. Ecommerce attribution platforms such as Triple Whale, Polar Analytics and Northbeam are built around Shopify and lead with MER. General BI tools such as Looker Studio, Domo and Klipfolio will compute anything you can model, at the cost of you owning the data modeling and its maintenance forever. Blended reporting tools, including ours, compute the blended layer out of the box from read-only connectors. If you would rather own the pipeline outright, pulling each platform's raw export into a warehouse through a data integration layer that connects APIs and databases gives you full control, and full responsibility for schema changes when a platform revises its API.

How do agencies show clients one clear performance number when running campaigns on Meta, Google and Amazon together?

The agencies that handle this well changed what they lead with rather than finding a better attribution engine. The client deck opens with blended efficiency across the whole account, because that is the only figure the client can check against their own bank balance. Channel detail follows as supporting evidence, explicitly labeled as platform-reported and explicitly not additive.

In practice the monthly report has a fixed shape. One headline efficiency number covering all spend and all revenue. Beneath it, the two revenue pools reported separately, so the client sees direct-to-consumer performance and Amazon performance without either contaminating the other. Beneath that, channel ROAS with a standing footnote explaining why the channel figures exceed total revenue. The footnote matters more than it sounds, because the question gets asked every single month and answering it once in writing buys enormous credibility.

The part agencies underestimate is the conversation at the start of the engagement rather than the reporting at the end of the month. A client who has been shown summed platform ROAS by a previous agency has an anchored expectation of a number that was never real, and the first honest report looks like a performance drop. Setting that expectation during onboarding costs one uncomfortable call and saves the account. If you are pricing the tooling side of this, we compared what the agency reporting platforms cost per client at 10, 25 and 50 clients in agency reporting software pricing.

Our Shopify revenue and our Meta Ads Manager ROAS do not match. Is this an attribution problem and how do I fix it?

It is an attribution problem, and the fix is not to close the gap but to make it stable and understood. Meta counts orders its pixel could tie to an ad interaction inside its attribution window, which includes view-through credit for people who never clicked. It also misses buyers whose tracking was blocked by browser restrictions, ad blockers or consent choices. Shopify counts every order once, regardless of how anyone arrived. Those two systems are answering different questions, so a gap is the expected outcome.

Industry reporting commonly puts the gap between platform-reported and analytics-reported conversions at 20 to 50 percent. A gap inside that band is normal. What deserves investigation is a gap that shifts sharply week over week with no matching change in how you are buying media.

One specific thing to check first if your Meta ROAS fell in early 2026 and you have been hunting for a creative problem since: on 12 January 2026 Meta removed the 7-day view and 28-day view attribution windows from the Ads Insights API, along with every combined window built on them. Long view-through credit simply stopped existing. Campaigns leaning on upper-funnel video were hit hardest, and many advertisers saw reported conversions drop with no change to targeting or budget. You may be comparing two different measurement regimes rather than two different creative strategies.

What to do this week

Four steps, in order, and none of them require new software to start.

First, pick your source of truth for revenue and write it down. For the direct-to-consumer pool it is Shopify or Stripe, because that is money that actually settled. For Amazon it is Amazon's own sales reporting. Everything else is a signal, not a fact.

Second, stop summing platform-reported revenue anywhere in your reporting, including in the spreadsheet nobody admits to maintaining. Replace the summed row with total recorded revenue and a blended efficiency ratio.

Third, calculate your break-even MER from contribution margin, which is 1 divided by your margin. At a 40 percent contribution margin you break even at an MER of 2.5. Without that line drawn, an efficiency ratio is a number without a verdict attached.

Fourth, chart blended CAC next to paid CAC. When they diverge you learn something no single metric tells you: blended falling while paid rises means organic is quietly absorbing the same spend across more customers, and raising the ad budget on the strength of the blended figure would be a mistake. Further reading on the ratios themselves is in MER benchmarks and blended ROAS reporting.

Where we fit, and where we do not

Worth being direct, because it saves everyone time. MixedMetrics connects Shopify, Stripe, Google Ads, Meta, TikTok, GA4, Search Console and Klaviyo read-only and computes MER, blended CAC, paid CAC and LTV:CAC on one board for a flat $79 a month. We do not connect Amazon Ads, and we also do not connect Microsoft Ads or LinkedIn Ads.

So for the exact scenario in the title, we solve the Meta and Google half properly and leave Amazon on its own line, which as argued above is the structurally correct answer anyway rather than a limitation we are dressing up. But if the majority of your $150k sits inside Amazon, a marketplace-native platform will serve you better and you should start there. We are also attribution-lite rather than a full multi-touch or media mix modeling product, so if you are at the spend level where incrementality testing and geo holdouts are on the table, that is a different category of tool. The broader setup for the channels we do cover is in cross-channel marketing analytics.