obOpen Banking Lab
GlossarySTETBlogContact
Technical

Cash flow underwriting

Cash flow underwriting (flow-based underwriting)

A credit-granting method based on analysing actual bank flows (AIS) rather than payslips and credit bureaus alone. It makes it possible to score poorly covered profiles (freelancers, first-time borrowers, SMEs).

See also:Scoring crédit alternatifAgrégation bancaireAISPEmbedded financeBNPL

Definition

Cash flow underwriting bases the credit decision on analysing the applicant's actual bank flows — recurring income, fixed expenses, average balances, volatility, stress signals — rather than payslips or the credit bureau.

It relies on bank account aggregation (AIS) and complements, often replaces, traditional bureau scoring. It is the major credit innovation of the 2020s in Europe and the United States.

What the bureau does not see

Classic bureau scoring relies on payslips, the tax notice, credit history (FICP, Schufa, FICO) and self-declared data. That model works poorly for:

  • freelancers and the self-employed (variable income, no payslip);
  • first-time borrowers (no history);
  • gig workers (couriers, ride-hail drivers, creators);
  • young SMEs (fewer than two years of financial statements);
  • recent migrants (no local history).

Cash flow underwriting looks at what actually happens on the account: income received, expenses paid, balances maintained.

The key metrics

Over 6 to 24 months of AIS history, the algorithm computes, among other things:

  • Recurring monthly income (RMI): the average of inflows identified as salary, benefits or client payments.
  • Income volatility: standard deviation, presence of months with no income.
  • Fixed expenses: rent, utilities, subscriptions, ongoing instalments.
  • Repayment capacity: RMI minus fixed expenses.
  • DTI (Debt-to-Income): credit instalments over income (target < 33% in France per the HCSF).
  • Balances: average and minimum over 6 months.
  • Overdrafts: frequency, duration, limits reached.
  • Stress signals: rejected direct debits, urgent transfers, gambling.
  • Expense categorisation: constrained vs discretionary share.

Three families of use cases

  1. Digital consumer credit (Younited, FLOA, Klarna, Alma): a decision in a few seconds at checkout.
  2. SME / freelancer business credit (Karmen, Defacto, October): opens credit to poorly served profiles.
  3. Salary advance / EWA (Rosaly, Stairwage, Bling): advances the current salary based on observed flows.

The technical integration

  1. The applicant starts their credit request.
  2. They consent to connecting their bank via an AISP (Bridge, Tink, Powens, Plaid).
  3. The AISP retrieves 12 to 24 months of flows and balances.
  4. A scoring engine (Algoan, Heron Data, or in-house) computes the features and applies the ML model.
  5. A decision in 2 to 10 seconds: approved or declined, rate, amount.
  6. If approved, disbursement on D+0 or D+1.

Without this setup, the same journey would take 3 to 10 days (collecting documents, manual review).

Cash flow underwriting vs alternative scoring

Close but distinct terms: alternative credit scoring is the general term (any scoring that is not pure bureau — digital behaviour, telecom, etc.), whereas cash flow underwriting is its dominant form in Europe since PSD2, specifically using AIS bank flows. One is a subset of the other.

What cash flow underwriting is not

  • Not a full substitute for the bureau: consulting the FICP remains mandatory for consumer credit (the Lagarde law). Cash flow complements, it does not replace.
  • Not immediate without consent: the applicant must consent to AIS, a friction point that costs 5 to 15% of drop-off in the funnel.
  • Not relevant for every credit: on a €300K mortgage over 25 years, wealth, savings and insurance matter more. Cash flow shines on small and medium amounts, over short durations.
  • Not outside the AI Act: from 2026, the AI Act classifies credit scoring as a high-risk AI system, with stronger requirements for data quality, governance and transparency.

In the PSD2 ecosystem

Cash flow underwriting is one of the most visible and value-creating use cases of AIS. Without PSD2, there would be no Younited, no European BNPL, no Karmen or Defacto, no credit embedded finance. It is also a growth driver for aggregators (Bridge, Tink, Plaid), which monetise their API through scoring.

Real-world examples

  • B2C consumer credit: Younited (€3K to €10K in 3 minutes), FLOA, Cofidis Connect, Cetelem Express.
  • BNPL: Klarna, Alma, FLOA Pay, Scalapay, Sequra score at checkout to decide on interest-free pay-in-3/4.
  • B2B SME / freelancer: Karmen (invoice financing), Defacto (short-term financing), October (SME loans), Mansa (freelancers).
  • Payroll EWA: Rosaly, Stairwage and Bling analyse the account to advance 30 to 50% of monthly net pay before payday.
  • Scoring building blocks: Algoan (FR), Heron Data (UK, B2B), Plaid Income/Assets (US), Bud Underwrite (UK).
  • Karmen case: a freelancer requests €5K to absorb a payment delay; Karmen aggregates the business account via Bridge, sees regular client payments and approves in 5 minutes at 7–12% annualised.
  • Performance vs bureau: on poorly covered segments, cash flow underwriting allows 20 to 40% more applicants to be approved at an equivalent default rate.
  • Outlook: with FIDA (2027+), it will extend to credit outstandings, savings and insurance — a 360° view that could add +5 to +15 Gini points to the model.

Sources

Back to glossary