Step conversion between each stage, plus end-to-end registration to first trade.
Fully-loaded acquisition cost per new first-trade customer. Marketing spend is an input you supply per market per month.
JP, KR and TW share one CPA. Spend is reported for JKT as a region rather than per country, so the region total is split by NFC share — which returns the region's own CPA to each of the three exactly. Their funnels are measured separately.
Total budget spent in the period on screen. The selected-month view shows that month alone; Year to date sums January through the selected month.
A partial period includes whatever spend has been booked so far. CPA can still be blank for that period because it excludes incomplete spend and its matching NFC denominator.
Everything the clients acquired during 2026 have billed so far. Not the market's whole book, and not one month's cohort in its first month. A month's figure is what all of those cohorts booked in that month, so the figures sum: Year to date gives the Jan-to-date total per market.
It grows through the year as earlier cohorts keep trading — most of a late month's number comes from clients won earlier, not that month's new ones.
New first-trade customers — the customers who completed their first transaction in the month. It is the acquisition denominator for CPA and revenue per customer, and the volume line the funnel is trying to move.
The UK is the mature-market yardstick and nothing else — measuring a younger market against a mature one reports its age, not its health. Growth markets are held to an agreed $450 CPA target; emerging markets, and every tier's funnel, are scored against the best achievement so far this year within their own tier. That bar is cumulative on purpose — year-to-date through the month on screen, not that month alone, so it does not swing on one month's noise.
Gaps are shown in percentage points: both sides are already percentages, so 44.9% against a 67.1% benchmark is 22.2pp behind — not "−33%", which is true of the ratio but reads as a third of registrations lost.
A target_cpa or
target_reg_to_nfc_pct in assumptions.csv beats the tier
benchmark. The map tooltip names which one a market's CPA was measured against.
Weighted 0–100 blend of the four metrics above. Each component scores 100 when it hits its benchmark; components with no data are dropped and the rest re-weighted, so a market is never penalised for data you have not supplied yet. Growth components are skipped for partial periods rather than comparing a part month against a full one.
CAC payback and LTV are deliberately absent: both rest on a gross margin, an active-customer count and a horizon that are not formally agreed, and an unagreed assumption reads as fact once it is on a dashboard.