Luca Barberis

When does each user cohort pay back its acquisition cost through transactions, interchange, and float, and how does it compound?

№ 111 · User ROAS curve

01

Definition

MetricCumulative gross profit per acquired user as a multiple of acquisition cost, by cohort month

Unitcumulative gross profit divided by CAC (x), by months since acquisition

Cumulative revenue per user (interchange + interest + fees) against user CAC by cohort. Build by aggregating per-cohort revenue streams against CAC. Different tiers have different curves; compare. Example: Tier 1 users 1.0x at M3, 3.4x at M24; Tier 3 1.0x at M12, plateaus at 1.6x.

ROAS
Return on ad spend. Revenue attributed to a campaign divided by the spend on that campaign.
CAC
Customer acquisition cost. Fully loaded sales and marketing spend in a period divided by new customers won in that period.
02

Benchmarks

Bottom 30%MedianTop 30%
1.0x not reached inside 24 months, or a curve that flattens below 1.0x because cohort revenue decays faster than it compounds1.0x reached at about 18 to 21 months, which is the five to seven quarter payback band disclosed by the largest US neobank1.0x reached within 12 to 15 months, on a fully loaded CAC that includes KYC, card issuance and sign-up incentives

Consumer fintech acquiring on paid and referral channels, cumulative cohort transaction profit against cohort acquisition cost. US and European consumer fintech, 2022 to 2026 cohorts. Payback defined as the point at which cumulative cohort transaction profit meets that cohort's acquisition cost. · The one clean disclosed anchor is Chime, which reported roughly seven quarters of payback for 2022 cohorts improving to five to six quarters by Q1 2026, on acquisition cost of USD 93 to USD 130 per member. That is a single company with an unusually strong referral mix, roughly half of new members, so it flatters the median. The definitional trap is CAC scope: a headline USD 50 neobank CAC becomes USD 269 or more per activated customer once KYC, sign-up bonuses, card issuance and funnel failure are included, which moves payback by a factor of five. State whether the chart uses signup CAC or funded-account CAC. Geography matters too, with Southeast Asian acquisition costs running 40 to 60% below North American and LatAm about 60% below, so a global blended curve mixes incompatible cohorts.

By category
CategoryBottom 30%MedianTop 30%
Interchange-led US neobank, disclosed cohort paybacknot separately disclosedabout 7 quarters for 2022 cohorts5 to 6 quarters
Credit-led LatAm digital bank, revenue ramp rather than paybackearly cohorts near USD 5 per monthblended ARPAC USD 15 against cost to serve USD 0.80 per monthARPAC of about USD 26 to 27 for cohorts of 5 to 8 years
03

When it looks bad

The cumulative curves bend over and go horizontal below the 1.0x line, so instead of continuing to climb toward and through payback they asymptote short of it, meaning cohort revenue is decaying at roughly the rate new revenue arrives.

The month-24 point sits at 0.71x for the 2024 cohort and 0.63x for the 2025 cohort, both flat since month 15. Every acquired user is a permanent loss of about a third of CAC, and scaling spend scales the loss.

04

What to do about it

  • Rebuild the chart on fully loaded CAC per funded account before making any spend decision. Including KYC at USD 5 to 15, sign-up incentives at USD 25 to 250, card issuance at USD 5 to 15 and activation drag can take a USD 50 headline CAC to USD 269 or more, and a curve drawn on the headline number will show payback that does not exist (Webtonic FoundryCRO 2026 benchmarks).
  • Attack the numerator through product depth rather than the denominator through cheaper traffic. Chime's payback moved from about seven quarters to five to six on early engagement work rather than on media efficiency, and members using six or more products generated 1.8x the average revenue while referring 1.6x as many friends (Chime Q1 2026 call and S-1/A).
  • Split the curve by channel and kill any channel whose cohort has not crossed 0.5x by month 12, since a channel that is behind at month 12 essentially never catches up once the retention curve has flattened.
  • Filter install and signup fraud before the cohort is drawn, because in Western European finance nearly one in two investment app installs was flagged fraudulent, which inflates the denominator, contaminates the retention base and produces a ROAS curve that is confidently wrong (AppsFlyer with Sensor Tower, 2026).
05

Sources

  1. Chime Financial (SEC Form S-1/A)Chime Financial, Inc. Form S-1/A · 2025 · n=1 platform, 8.6m active members at filing Defines payback as the point at which cumulative transaction profit for a member cohort meets the acquisition costs incurred when that cohort first became active, and defines a primary relationship as 15 or more card purchases per month or a qualifying USD 200 direct deposit. sec.gov ↗
  2. Chime (Q1 2026 earnings call transcript, The Motley Fool)Chime (CHYM) Q1 2026 Earnings Call Transcript · 2026 · n=1 platform, post-IPO reporting Early engagement initiatives lowered acquisition costs and brought the cohort payback period to five to six quarters. fool.com ↗
  3. The Free Toaster, analysing the Chime S-1Chime S1: Why Their Flywheel is So Fly · 2025 · Chime S-1 disclosures, 2022 to 2024 cohorts Acquisition cost per new active member of USD 109 in 2024, or USD 91 excluding broader brand marketing, with payback in about seven quarters for 2022 cohorts and faster for the Q1 and Q2 2023 cohorts. thefreetoaster.com ↗
  4. Bob Hammel (Substack), analysing Chime disclosuresChime Financial · 2026 · Chime quarterly disclosures 2022 to Q1 2026 Quarterly acquisition cost per member ranged from USD 93 to USD 130 since 2022 and was USD 107 in Q1 2026; LTV to CAC of 8x or better implies payback of five to six quarters, improved from about seven a year earlier. bobhammel.substack.com ↗
  5. Nu Holdings (Nubank)Nu Holdings Ltd. Reports Fourth Quarter and Full Year 2025 Financial Results · 2026 · n=1 platform, 131m customers, record quarterly net income of USD 895m ARPAC of USD 15.0 and cost to serve of USD 0.80 per active customer per month, with an efficiency ratio of 19.9% and ROE of 33%; 15 to 90 day NPL declined to 4.1%. international.nubank.com.br ↗
  6. Nu Holdings (Nubank)Nu Holdings Ltd. Reports First Quarter 2025 Financial Results · 2025 · n=1 platform, LatAm consumer base above 118m customers ARPAC of USD 11.2 with mature cohorts at almost USD 26, monthly activity rate of 83.2% and cost to serve of USD 0.70. international.nubank.com.br ↗
  7. Webtonic (FoundryCRO benchmark set)Fintech Digital Marketing Statistics: 2026 Performance Report · 2026 · Compiled 2026 fintech marketing benchmarks across 26,000+ active fintechs; underlying sample not disclosed A neobank reporting USD 50 CAC pays USD 269 or more per activated customer once KYC at USD 5 to 15, sign-up bonuses at USD 25 to 250, card issuance at USD 5 to 15 and activation-rate drag are included; fintech retention falls from 28% at day 1 to 14% at day 30. webtonic.io ↗
  8. DatabrainFintech KPIs and Metrics: 12 Indicators and Formulas · 2026 · Compiled fintech operating benchmarks, sample not disclosed Activation benchmarks of 30 to 60% for neobank first-funded account, 50 to 80% for payments first transaction and 25 to 50% for wealth-tech first funded portfolio; a USD 40 signup CAC becomes USD 80 to 120 per funded customer once funnel failure is included. usedatabrain.com ↗
  9. AppsFlyer, with Sensor Tower and Google AdsWhere European Finance Apps Win or Lose Users (State of Finance for Marketers: Europe 2026) · 2026 · n=approximately 300 European finance apps, 2.4bn installs and USD 1bn of UA and remarketing spend, Q2 2024 to Q1 2026 Web-to-app drove 41.8% of owned media conversions in Western European finance apps, nearly one in two Western European investment app installs was flagged fraudulent, and BNPL installs grew 40% while crypto fell 35%. appsflyer.com ↗
  10. LTV CAC Book, citing FirstPageSage, Benchmarkit and KeyBancCAC Benchmarks 2026: Customer Acquisition Cost by Industry and Stage · 2026 · FirstPageSage 120+ firms, Benchmarkit approximately 1,000 companies, KeyBanc SaaS Survey 2025 to 2026 Primarily North American figures, with Southeast Asia running 40 to 60% lower, Latin America about 60% lower and Southern and Eastern European CPCs 2 to 5x lower than US and UK equivalents. ltvcacbook.com ↗