- 2 counts: platform customers and finance customers, same period.
- Under 10% is acceptable, above 20% serious (Usermaven, 2026).
- 31%: 140 platform customers against 96 in finance.
How do you calculate attribution variance?
Attribution variance is platform-reported customers minus finance-counted customers, divided by platform-reported customers, for the same period.
Count customers rather than conversions, under one definition and one window.
What is a healthy attribution variance?
Under 10% reads as acceptable, 10% to 20% as worth investigating and above 20% as serious (Usermaven, published 20 May 2026, updated 18 September 2026).
| Band | Variance | Action |
|---|---|---|
| Acceptable | Under 10% | Publish the number and keep the weekly review |
| Investigate | 10% to 20% | Check windows and duplicate events first |
| Serious | Above 20% | Match every row before the next budget increase |
A 12% gap opened weekly by a named person catches a broken tag inside seven days.
What does attribution variance look like on real numbers?
When a platform reports 140 customers and finance counts 96, the attribution variance is 31%.
| Source | Customers reported | Variance |
|---|---|---|
| Ad platforms | 140 | |
| Finance | 96 | |
| Attribution variance | 44 | 31% (44 divided by 140) |
On $100,000 a month of spend, cost per customer moves from $714 to $1,042, and one of those decided your budget, didn't it? What did a customer cost you last month, without a caveat?
Which seam does attribution variance measure, and who owns it?
Attribution variance measures seam 01, SPEND to ATTENTION, the handoff between your media buyer and your finance lead.
It is a revenue seam because the buyer reads the platforms and finance reads the bank, and this rate compares the two, the reasoning in why platform and finance customer counts never match. Coach Kyle's 30 August 2026 launch repaired 20 of 57 broken tracking messages first, then drew 412 registered, 142 live, 37 applications, 30 calls booked and 4 closes at $6,800, which is why you audit tracking before a launch. Read beside yield per subscriber and committed to appeared rates, the gap becomes one reconciled count, with an AI employee pulling both exports each Monday for the person who owns it, so who opens it first?
Forty five minutes sets your platform count beside your finance count, in dollars.
Click the button below.
Answer a few short questions.
Pick a time on Jon's calendar.
What else do people ask about attribution variance?
Why does Meta report more purchases than my store?
Meta defaults to a 7 day click and 1 day view window (Usermaven, 2026), so one customer can be claimed by two platforms and counted twice.
How often should marketing and finance reconcile?
Weekly for the headline gap, monthly for the full variance list, with one named person opening the number every Monday.
Is attribution variance the same as the attribution gap?
Yes. The pillar calls it the attribution gap; the glossary calls it attribution variance. Both are platform-reported customers minus finance-counted customers, divided by platform-reported customers.