Your ad platform says it drove 92 conversions last week. Your store logged 68 orders. Your analytics tool reports a third number. None of them are wrong — the problem isn't the data. It's expecting three measurement systems, each built to serve its own product, to agree by default.
Three numbers, each technically "correct"
The gap isn't a malfunction. Every platform counts the same sale in the way that flatters its own story.
Attribution windows differ
Seven-day click, one-day view, linear, data-driven — the same order gets claimed by different ads under different models. Longer windows mean more conversions claimed, and better-looking reports.
Conversion definitions differ
Order placed, payment settled, session-deduplicated: three moments, three counts. The ad platform may record "orders after a click," the store records completed payments, and the analytics tool deduplicates by session.
Time and identity differ
Time zones, report-day boundaries, cross-device identity merging — each alone moves the total by a few percentage points. Stack them and you get the reconciliation that never ends.
Attribution gaps aren't data errors — they're each platform telling the story that favors itself.
How to reconcile without going mad
Mature teams don't try to eliminate the gap. They manage it:
- Define before you compare. Document what every metric means; one number, one meaning, everywhere.
- Pick one source of truth. Treat the rest as reference and log the deltas instead of re-litigating them weekly.
- Explain the gap, don't chase it. Turn "why are we 24 orders apart" into documented knowledge, not a recurring surprise.
Aligning decisions matters more than aligning numbers
The real cost of disagreeing numbers isn't inaccuracy — it's drained trust. When every channel can produce a flattering chart, budget meetings turn into people talking past each other. Fix the definitions and the source of truth, and the conversation moves from "which number is right" to "what do we do first."
That is also where automated aggregation beats hand-built spreadsheets: a machine applies the same rules every day, so definitions stop drifting with whoever built the sheet. Diagnosis on a consistent baseline gives conclusions you can compare — and repeat.
Key takeaways
- Attribution windows and conversion definitions come before any reconciliation.
- One set of definitions, one source of truth, deltas logged.
- Swap the "which number is right" debate for "what do we do first."