VALIDATED AUC-ROC 0.860 · 47 ENGINEERED FEATURES · FAIR LENDING COMPLIANT

Credit scoring that
sees cash flow

Assess small business creditworthiness from bank transactions alone. No financial statements. No credit history. Just the cash flow truth.

RESEARCH CALIBRATED AGAINST JPMorgan Chase Institute Federal Reserve SBA
0
AUC-ROC
Predictive accuracy
0
Features
Engineered from transactions
0
Default Reduction
vs. unscreened lending
0
Seconds
Application to decision

Everything you need to lend smarter

Built on peer-reviewed research. Designed for credit unions and community banks.

Instant Scoring

Upload bank transactions and receive a credit decision in seconds. Our ML pipeline extracts 47 features and scores in real-time.

🔬

5-Category Feature Engine

Revenue patterns, expense behavior, cash position, behavioral signals, and derived ratios — each capturing distinct dimensions of creditworthiness.

⚖️

Fair Lending Built In

Four-Fifths Rule monitoring, no demographic inputs, proxy variable testing. ECOA-compliant adverse action codes generated automatically.

🎯

Explainable Decisions

Every score includes the top factors driving the decision — positive and negative — mapped to plain-English reasons your loan officers understand.

📊

Portfolio Analytics

Real-time dashboards showing risk tier distribution, origination trends, approval rates, and early warning signals across your entire book.

🔌

API-First Architecture

REST API with sub-200ms response times. Integrate with your LOS, core banking system, or member-facing application in days, not months.

From transactions to decisions

01

Connect

Business links bank accounts via Plaid or uploads 12–18 months of transaction data (CSV, OFX, QFX).

02

Extract

Our pipeline categorizes transactions and extracts 47 features across revenue, expenses, cash position, behavior, and derived ratios.

03

Score

XGBoost ensemble model scores the business on a 0–100 scale. Probability-calibrated output enables risk-based pricing.

04

Decide

Loan officer receives score, risk tier, key factors, recommended terms, and ECOA-compliant adverse action codes if declined.

Numbers that matter

Impact on Lending Operations

Underwriting Cost$3,500$75-98%
Processing Time30 days2 seconds-99%
Approval Rate15-20%65-80%+3.5×
Portfolio DefaultUnknown10-14%Managed
Loan Volume$18.3M$50M++173%

Model Validation (4 Algorithms)

XGBoost0.860
Random Forest0.860
Logistic Regression0.861
Neural Network0.862
Key Insight

All four models converge at AUC ~0.86, proving the signal lives in the features, not the algorithm. Even simple logistic regression matches complex ensembles.

Start with a free pilot

No commitment. Validate on your own data before scaling.

Pilot
Free90 days

Validate with your own data

Up to 500 scores/month
Dashboard & analytics
Single user
Email support
CSV upload only
Start Pilot
Growth
$2,500/month

For active lending programs

Up to 5,000 scores/month
Full API access
5 users
Fair lending reports
Plaid integration
Priority support
Get Started
Enterprise
Custom

For large portfolios

Unlimited scores
Dedicated infrastructure
Unlimited users
Custom model training
On-premise option
Dedicated CSM
SLA guarantee
Contact Sales

Built on science, not hype

CashFlow IQ implements methodology from peer-reviewed research on cash flow-based credit scoring, calibrated against data from 600,000+ real businesses.

"Cash Flow-Based Credit Scoring for Small Business Lending: A Machine Learning Approach Using Bank Transaction Data"
Naga Rakesh Chinta — Indian Institute of Technology Kanpur
DATASET

10,000 businesses 5.2M transactions

BENCHMARKS

JPMorgan Chase Inst. Fed Reserve SBCS SBA 7(a) FOIA

VALIDATION

5-fold cross-validation Fair lending tested Calibration verified

Ready to close the credit gap?

Join the community lenders bringing cash flow intelligence to small business lending.