Key implication: The centre of gravity in business lending is shifting to fast, data-rich, embedded credit tailored to small and medium enterprises. In Australia, the SME lending pool sits in the $450–$631 billion range (circa 35% of total business lending), with a pre-2024 CAGR near 9%. Globally, McKinsey pegs the annual SME financing shortfall at roughly US$5.2 trillion. The prize is real; so are the execution risks.

Market context: A big pool with uneven access

SMEs comprise the bulk of employer firms and are disproportionately credit-constrained—especially newer, asset-light and service-based businesses. Australia mirrors global patterns: strong aggregate volumes, patchy access, and cyclical headwinds that delay capex even as working-capital needs rise. The funding gap persists because legacy scorecards struggle with thin-file borrowers, manual processes inflate cost-to-serve, and collateral-light lending carries higher perceived risk.

Two growth vectors are changing the math. First, digital adoption among SMEs (cloud accounting, e-invoicing, e-commerce, real-time payments) creates richer, more verifiable data. Second, the Consumer Data Right (open banking) is maturing, enabling consented access to bank transaction data. Blend these with modern decisioning and time-to-yes drops from weeks to hours; cost-to-serve falls; risk models improve.

Competitive dynamics: Banks, fintechs and platforms converge

Porter’s lens clarifies the scramble:

• Threat of substitutes: Non-bank fintechs offer 24–48 hour decisions and working capital facilities aligned to sales cycles—compelling alternatives to bank overdrafts.
• Bargaining power of buyers: SMEs now compare across brokers, fintechs and their existing platforms (payments, accounting, marketplaces). Speed, transparency and repayment flexibility outweigh headline rate for many short-duration needs.
• New entrants: Embedded finance is the wildcard. E-commerce, payments and SaaS platforms with privileged data pipes (think global examples like Shopify Capital and Amazon Lending) can price risk with transaction-level telemetry and disburse inside the workflow.
• Industry rivalry: Banks defend with balance sheet strength, lower funding costs and trust; fintechs counter with product agility and UX. Partnerships and white-label models are proliferating.

Result: Share migrates to whoever controls the data, the moment of need and the user experience. Brokers remain vital but must operate as orchestrators across multiple lenders and data rails.

Technical deep dive: From collateral to cashflow models

Modern SME underwriting fuses multiple consented feeds to solve the thin-file problem:

• Bank transactions (via open banking): cashflow volatility, seasonality, concentration risk.
• Accounting ledgers (Xero, MYOB, QuickBooks): receivables ageing, payroll cadence, tax liabilities.
• Commerce and payments (POS, gateways, marketplaces): sales velocity, refund rates, basket size, cohort performance.
• E-invoicing and e‑procurement: invoice authenticity, buyer quality, payment time.

Machine learning models exploit these signals to estimate probability of default, expected loss and prepayment, while rule engines enforce policy constraints (industry blacklists, AML/KYC flags, related-party checks). Mature lenders bake in fraud controls (synthetic identity detection, document forgery analysis) and model risk management (champion–challenger, backtesting, bias monitoring). Done well, this reduces manual touches by 30–50%, improves early‑stage conversion, and tightens loss distribution.

Case examples: What’s working—and why

OnDeck (Australia) built a strong niche by ingesting bank transaction data and accounting feeds to underwrite unsecured loans rapidly—competing on time-to-funds and transparency rather than branch presence.

Prospa scaled with ML-driven decisioning and a broker-first distribution engine, offering business loans and revolving lines aligned to SME cash cycles, with near-real-time drawdowns.

Judo Bank took a hybrid route: relationship bankers supported by modern platforms to speed complex SME deals—showing that technology augments, not replaces, human credit judgement for larger exposures.

Global benchmarks such as Shopify Capital, Amazon Lending and India’s NeoGrowth demonstrate the power of embedded models: native data, push-button offers inside existing workflows, dynamic repayments tied to sales. The common thread is privileged data plus instant distribution.

Broker strategy: From product placement to capital orchestration

For brokers diversifying beyond mortgages, SME credit is less about a single product and more about an ongoing working-capital relationship. A practical playbook:

• Segment and prioritise: Identify existing mortgage clients with ABNs; triage by industry volatility, seasonality and data readiness (cloud accounting adoption is a green flag).
• Curate a multi-lender panel: Combine bank overdrafts for rate-sensitive needs with fintech lines for speed/flexibility, plus asset finance for equipment-heavy clients.
• Lean on data: Introduce open banking and accounting-data consent as standard. Clients trade data for speed and optionality; brokers gain stronger files and higher first-time approval rates.
• Build recurring value: Lines of credit and embedded working-capital products create repeatable, annuity-like commissions, smoothing broker revenue across cycles.
• Compliance-by-design: Standardise AML/KYC, privacy and consent flows; document suitability and data provenance to satisfy lenders and regulators.

The economics can be compelling: higher margins than prime mortgages, faster sales cycles and stickier relationships—balanced against higher diligence and portfolio monitoring.

Execution reality: Risks, controls and operating model

This is not a free lunch. Three realities to manage:

• Credit cyclicality: SME defaults can spike faster than consumer. Build early-warning systems using live transaction and ledger data (missed BAS, supplier stretch, payroll anomalies).
• Fraud and first-party misuse: Deploy layered identity checks, document detection, device fingerprinting and post-funding monitoring (cash diversion flags).
• Legacy constraints: Banks wrestle with core integration and policy rigidity; fintechs face funding-cost volatility. Both should adopt API-first orchestration, separate policy from code, and stand up dedicated SME squads with credit, data science and risk ops under one roof.

Funding strategy matters. Banks leverage low-cost deposits; fintechs diversify through warehouse lines and ABS once scale and performance data allow. A disciplined path from balance-sheet lending to capital markets can reduce cost of funds by 100–200 bps over time if loss performance holds.

Regulation and data: The next unlocks

Open banking under Australia’s Consumer Data Right continues to expand, improving data quality and consent mechanics. Expect broader business-data availability and easier multi-institution connectivity, enabling richer underwriting and portability of credit profiles. Globally, regulators are encouraging competition and transparency while tightening on fraud, privacy and responsible lending—raising the bar for data governance and model risk management.

Outlook: Embedded, event-driven and AI-assisted

The next phase is event-driven finance: pre-approved limits surfaced when the SME runs payroll, issues invoices or hits a large order. Generative AI will accelerate broker and lender workflows—drafting credit memos, triaging cases, and explaining adverse decisions in plain language—while traditional ML continues to do the heavy lifting on risk. Winners will combine three assets: proprietary or privileged data access, instant distribution via trusted channels, and industrial-grade risk controls.

The takeaway for Australian players is pragmatic: chase speed without abandoning prudence, partner where platforms already own the relationship, and operationalise data consent as a feature—not an afterthought. The gap is large, the market is ready, and first movers are already compounding advantages.