By Ryan Richardson · Published 8 October 2026
Grading on settled cash is step 53 (Part IX, Read it and decide), marked hard difficulty, almost nobody does this. It is the single rule everything else in this part of the matrix depends on.
Platform reporting showed $0.06 return per dollar on A$120.70 of spend. The ledger, pulled from settled charges, showed A$263.94 across 7 settled sales, a return of 1.55. The ad set was about to be killed on the platform's number. Nothing is killed or scaled on platform reporting alone after this; the ledger decides.
Settled cash is the gross charge, minus refunds, disputed amounts and fees, in one currency, converted at the transaction rate rather than today's rate. Declare one system as the ledger of record, usually the payment processor's own settlement record, and declare a settlement lag: refunds on a cheap digital product cluster in the first fourteen days, so a week's figure isn't final until about day fourteen after that week closes.
Pull settled charges for the period, read the campaign parameter from each charge's metadata, group by campaign; that is revenue per campaign, and the only version of it that was never asked to grade its own work. Read the last seven complete days, excluding today, because today is always partial. Compare cohort to cohort, not week to week, since week-over-week mixes new buyers with delayed buyers from weeks earlier. Ban return on ad spend from the page entirely: it's a ratio of a modelled numerator to a real denominator and moves for reasons that have nothing to do with the business. Once a month, add one week's settled charges by hand against a bank statement, the only check here that doesn't depend on the thing it's checking.
Charges with no campaign parameter, which bias whatever bucket they get dropped into. Refunds counted as revenue. Currency conversion landing on the wrong date, enough to flip a decision on a thin margin.
| Claim | Value | Source |
|---|---|---|
| Ad set return, platform-reported vs bank-settled | $0.06/dollar reported on A$120.70 spend vs A$263.94 across 7 settled sales (return of 1.55) | Measured in Real Money, Field Manual |
| Settlement lag for refunds on a cheap digital product | refunds cluster in the first 14 days, so a week's figure isn't final until about day 14 after that week closes | The Sixty Steps manuscript |
| Rule for building the weekly settled-cash page | read the last 7 complete days excluding today; compare cohort to cohort, not week to week; ban return on ad spend from the page entirely; reconcile one week by hand against the bank statement once a month | The Sixty Steps manuscript |
| Matrix status for this step | Hard difficulty, almost nobody does this, automated tooling | The Sixty Steps matrix |