Three platforms all claimed the same sale.
A single order can land in Meta, Google and Klaviyo at once, each counting it in full. Add the dashboards up and you get more revenue than your store actually made. Here is why, and the two numbers that correct it.
Meta, Google and Klaviyo each count a conversion they touched, so a single sale can appear in all three dashboards at once. Summed, platform-reported revenue routinely exceeds what the store actually banked. The honest corrections are blended MER, which divides real ledger revenue by total spend and cannot double-count, and incrementality tests that measure what the ads actually caused.
Open Meta Ads Manager, Google Ads and Klaviyo on the same Monday morning and add up the revenue each one claims for the weekend. The total will almost always be larger than what your Shopify dashboard says you actually sold. Nobody is lying. Every platform is counting the same sale, and each one counts it in full.
This is not a bug in any single platform. It is what happens when three tools each measure the same journey through their own window and each take full credit. The result is an attribution overlap that inflates reported revenue above the truth, and it quietly distorts every budget decision built on top of it.
Why does the same sale show up three times?
Follow one buyer. On the weekend they see a Meta video ad and do not click. Two days later they search your brand name and click a Google ad. That evening they open a Klaviyo flow email, click through, and buy one A$120 order. Three platforms, one sale. Meta records a view-through conversion for the impression it served inside its window. Google records a last-click conversion. Klaviyo attributes the order to its email click. Each is following its own attribution model, and each books the whole A$120.
The windows make the overlap structural, not occasional. Meta's default counts 7-day click and 1-day view. Google counts up to 30-day click. Klaviyo attributes on a click and open window of its own. Any purchase that two of them touched inside those windows is recorded by both in full. Nobody double-counts on purpose; the platforms simply cannot see each other, so they cannot subtract. That is why a summed ROAS across channels is not a real ratio. Its numerator counts dollars that do not exist.
How much are the platforms over-counting?
Small at the order level, large at the account level. The measurement firm Measured documents cases where ad platforms collectively claim credit for up to 140% of actual revenue (Measured, 2024). Independent benchmarking from Varos puts Meta roughly 26% above third-party analytics and Google Ads 15% to 20% high once modelled conversions fill the gaps (Varos industry benchmark, 2024). Stack email attribution on top and the summed figure drifts further from the ledger.
Put numbers on it. The store below banks A$400,000 in a month, verified against its Shopify ledger. Its three platforms report A$500,000 between them. That A$100,000 gap is not extra revenue. It is the same orders counted more than once.
Divide the summed A$500,000 by that spend and you get a flattering 5.0. Divide the real A$400,000 that hit the bank by the same spend and you get 4.0. The first is fiction built on double-counting. The second is blended MER, and it is the one you can bank on.
What is blended MER, and why can it not double-count?
MER, the marketing efficiency ratio, is total revenue divided by total marketing spend. One numerator, one denominator, no per-platform claims involved. Because the numerator is the store's actual ledger revenue rather than a sum of dashboards, it is mathematically impossible for it to exceed 100% of sales. Whatever Meta, Google and Klaviyo argue about who caused what, they are dividing up a pie whose true size MER already knows.
MER has a break-even, just like ROAS. Break-even MER is 1 divided by contribution margin, the same inverse logic behind break-even ROAS: at a 25% contribution margin a store needs a blended 4.0 just to cover variable cost, before it has paid a cent toward acquisition. The demonstrative store above sits at exactly 4.0, so the honest read is that it is running at zero contribution after cost of goods. The summed 5.0 hid that completely.
This is the seam Blufire works in. It reconciles the revenue each platform reports against the store's actual ledger, so phantom dollars never enter the numbers, and it reads the whole store in contribution margin rather than revenue. What comes out is one blended, margin-true efficiency figure instead of three dashboards competing to claim the same order.
Does a clean MER prove the ads caused the sales?
No, and this is the second correction. MER removes the double-counting, but a healthy MER can still be built on demand you already owned. Branded search is the classic case. Someone who has already decided to buy types your name, clicks the ad, and the platform books a conversion it did not create. Turn the ad off and most of those clicks reappear as free organic results.
That is incrementality, and it has a settled experimental answer. Economists at eBay switched paid search off in some regions and left it on as a control. Almost all the forgone paid clicks were immediately recaptured by organic listings, and average measured returns on non-brand terms were negative (Blake, Nosko & Tadelis, Econometrica, 2015). The lesson is not platform-specific. Any channel can report a strong number while adding little real revenue.
You cannot read incrementality off a dashboard. You have to withhold the spend and measure the difference. A holdout test suppresses a channel or audience for a matched group and compares outcomes. A geo-lift test does the same across regions, turning spend off in some markets and holding it steady in others, then reading the gap. Run one on your two highest-spend channels once or twice a year. Branded search and retargeting are where the phantom revenue concentrates.
Three platforms, one sale, three full claims. Stop summing dashboards that were never meant to be added. Report one blended MER against your real ledger revenue, hold it to the break-even your margin demands, and correct it for cause with a holdout once or twice a year. That is the honest version of the number, and it is the only one worth setting a budget on. The full method sits in The Math.
- Blake, T., Nosko, C. & Tadelis, S. (2015). "Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment." Econometrica 83(1). Peer-reviewed eBay holdout on branded and non-brand paid search incrementality.
- Measured (2024). Incrementality vs attribution vs MMM decision tree. Ad platforms collectively claiming up to 140% of actual revenue, and non-incremental ranges for branded search and retargeting.
- Varos industry benchmark (2024). Meta roughly 26% over-reported conversions versus third-party analytics; Google Ads 15% to 20% over-attribution under modelled conversions.
The table and chart marked "Example numbers" use illustrative figures, sized to the published over-report ranges above and applied to a typical store. They are not measured Blufire client results. The arithmetic is currency-neutral and applies to AUD reporting unchanged.