Attribution, autopsied

Three platforms. One order. 300% of the credit.

By the MoonSignal team · operators, not bloggers · Updated July 21, 2026

Watch one real customer: sees your Meta ad Tuesday (doesn't click). Gets your abandoned-cart email Thursday. Googles your brand Friday, clicks the search ad out of laziness, buys $80.

Meta claims it (1-day view / 7-day click). Google claims it (brand click). Klaviyo claims it (email engagement window). One $80 order, three dashboards each booking $80. Add the platforms' "attributed revenue" and you invent money that never existed — this is why every blended report you've ever summed felt too good.

The mechanisms doing the lying

Touch ≠ cause. Platforms credit conversions they touched inside a window, not conversions they caused. View-through is the boldest move: an impression scrolled past yesterday claims today's purchase. It exists because it makes reported ROAS bigger — turn 1-day-view off in your columns and watch "performance" deflate to something honest. Warm traffic is the easy con: brand search and retargeting mostly harvest people already coming — that branded campaign showing 20x isn't generating 20x; it's standing at the finish line taking photos. (Run the thought experiment: if you paused brand search for a week, how many of those buyers would have clicked the free organic result an inch below?)

MER: the number with no opinion

MER = total store revenue ÷ total ad spend, all platforms. No windows, no models, no self-grading — your bank account, as a ratio. Breakeven MER = 1 ÷ blended gross margin. 60% margin → 1.67. Above it, the whole machine prints; below it, you're buying revenue with your own money while three dashboards congratulate you.

The marginal trap that kills scalers

MER is an average, and averages hide the edge. Going $5k→$10k/month, the new $5k performs worse than the first (auction depth: you exhaust cheap buyers first). Blended MER 3.5 can conceal marginal MER 1.4 on the last dollars — profitable-looking growth, unprofitable growth dollars. When scaling, compute MER on the increment: (new revenue − old revenue) ÷ (new spend − old spend). That number decides whether to keep pushing, and almost nobody looks at it.

The working system

Weekly: MER vs breakeven — the profit verdict. Daily: platform metrics for relative comparison only (Campaign A vs B inside one platform — both inflated by the same lie, so the comparison survives). Monthly: divergence trend between platform-claimed revenue and store revenue — a sudden widening means tracking broke or attribution inflation grew. And when anyone — agency, dashboard, or algorithm — argues for scale, ask for the marginal MER. That question has saved more money in our teardowns than any bid strategy ever has.

Free · 15 minutes

Want this run on your actual account?

MoonSignal is a private AI media buyer inside Claude — it reads your real numbers, finds the leaks, and builds the fix. You approve every change.

Get a free live account teardown →