How OnDeck's AI Growth Data Proves Bank Verification Software for Funders Must Evolve

Platform Lenders Are Widening the Verification Gap

Square's Q2 earnings letter delivered a line that should keep every independent MCA funder awake at night: "Square Loans are underwritten on data banks can't see and serve sellers banks won't." That single statement captures the structural advantage platform lenders hold over independent funders. Square's loan cohorts have maintained loss rates below 4% through every economic cycle the company has faced. Shopify Capital, meanwhile, originated $1.4 billion in Q2 alone, with CFO Jeff Hoffmeister noting that "Capital was a larger driver this quarter" while loss rates stayed at normalized levels.

These numbers are not abstract. They represent a competitive reality for every independent MCA funder trying to grow in 2026. Platform lenders see every transaction flowing through their payment rails. They know daily sales volumes, seasonal patterns, refund rates, and chargeback frequencies in real time. Independent funders see bank statements, sometimes manipulated ones, and hope for the best.

The question is no longer whether bank verification software for funders needs to improve. The question is how fast it needs to evolve before the gap becomes permanent. OnDeck's continued AI-driven origination growth, the 2026 Inc 5000 list featuring several fast-growing MCA funders, and Block's consistent sub-4% loss rates all point in the same direction: the funders investing in verification technology are winning, and those relying on legacy processes are falling behind.

Why Static Bank Statement Verification Fails Against Platform Data

The Data Asymmetry Problem

When Square underwrites a loan, its machine learning models draw from millions of data points: card-present versus card-not-present ratios, average ticket sizes over 90-day windows, velocity of daily deposits, customer return rates, and dozens of other signals invisible to external lenders. Shopify Capital does the same with its merchants' order histories, fulfillment timelines, and payment processing patterns. This is first-party data at massive scale, and no PDF bank statement can replicate it.

Independent funders are stuck on the other side of this asymmetry. They receive bank statements that may or may not reflect reality. A fraudster can alter transaction descriptions, fabricate deposits, or present statements from a different business entirely. Even legitimate applicants sometimes submit outdated or incomplete documents. The static nature of document-based verification means underwriters are always working with a snapshot, never a live view.

AI Underwriting Is Only as Good as Its Inputs

OnDeck has invested heavily in AI-powered underwriting models. Their approach uses machine learning to analyze cash flow patterns, predict repayment probability, and automate approval decisions. But here is the part that often gets overlooked: AI underwriting models are only as reliable as the data feeding them. If the input is a manipulated bank statement or an incomplete transaction history, even the most sophisticated model will produce flawed output.

This is where the verification layer becomes critical. Before any AI model scores a deal, someone or something needs to confirm that the underlying financial data is authentic. As we explored in our analysis of how Square's sub-4% loss rates expose the bank verification software gap for funders, platform lenders solve this problem by never relying on external documents at all. Independent funders need a different solution.

Session-Based Verification Closes the Gap

The most effective alternative to first-party transaction data is session-based verification: watching an applicant navigate their actual banking portal in real time. This approach captures information that static documents cannot. An underwriter can see the URL bar confirming the banking domain, observe transaction histories loading dynamically, verify account holder names, and confirm that balances match stated figures. No PDF manipulation survives this level of scrutiny.

Exact Balance takes this concept further by making it asynchronous. Instead of scheduling a live video call and walking the applicant through their portal step by step, the applicant receives a secure link, records their banking session at their convenience, and submits the recording for review. An AI-guided floating coach walks them through each required step, verifying completion in real time. The underwriter reviews the recording later, on their own schedule, with a complete activity log and timestamp trail.

This approach gives independent funders something remarkably close to what platform lenders have: verified, visual evidence of real banking activity. It is not the same as owning the payment rails, but it is far more reliable than trusting a PDF.

What the Inc 5000 Reveals About Verification Infrastructure and Growth

The 2026 Inc 5000 list included several notable MCA funders. Specialty Capital ranked at number 165 with 1,974% three-year growth. Parafin came in at 357 with 969% growth. FundCanna appeared at 434. These numbers reflect origination velocity, but they also imply something about operational infrastructure. You do not grow originations by nearly 2,000% over three years without a verification process that can scale alongside volume.

Consider the math. If a funder processes 200 deals per month and relies on scheduled live verification calls, each call taking 20 to 30 minutes of underwriter time plus scheduling overhead, that funder hits a throughput ceiling quickly. Doubling origination volume means doubling verification staff, doubling scheduling complexity, and doubling the odds of timezone conflicts, missed appointments, and delayed funding decisions. As we detailed in our analysis of how Inc 5000 MCA growth rates expose the bank verification software gap, the fastest-growing funders are the ones investing in infrastructure that decouples verification capacity from headcount.

Asynchronous verification solves this scaling problem directly. When applicants record on their own time and underwriters review on theirs, there is no scheduling bottleneck. A single underwriter can review recordings at two to three times the speed of conducting live calls, because they can skip idle time, pause and rewind for closer inspection, and batch reviews during focused work blocks. The verification process becomes a content review workflow rather than a real-time coordination challenge.

This matters even more for Canadian funders dealing with cross-border applicants or multiple time zones. A funder based in Toronto processing deals from Vancouver to Halifax no longer needs to align schedules across a four-and-a-half-hour timezone spread. The applicant records when it suits them. The underwriter reviews when it suits the team.

How AI Fraud Detection Strengthens Session-Based Verification

The shift to session-based verification also opens the door to AI-powered fraud detection layers that static documents cannot support. When a bank verification recording captures a live session, machine learning models can analyze multiple fraud signals simultaneously.

Domain validation checks confirm that the banking portal URL matches a known financial institution. Visual consistency analysis flags recordings where page layouts, fonts, or transaction formatting deviate from expected patterns for a given bank. Behavioral anomaly detection identifies sessions where mouse movements suggest scripted or automated navigation rather than organic human interaction. Timestamp correlation verifies that the recording date matches the claimed transaction period.

These are not theoretical capabilities. They represent the direction that verification technology is moving in 2026, and they are only possible when the verification input is a rich, visual recording rather than a flat document. Exact Balance's platform captures these recordings with full audit trails, giving underwriters both the raw visual evidence and the structured metadata needed for AI-assisted review.

The Federal Reserve's small business lending data continues to show that alternative lending volumes are growing while traditional bank lending to small businesses remains constrained. As volume grows, so does fraud sophistication. The funders that survive and thrive will be those whose verification technology evolves faster than the fraud techniques targeting them.

Frequently Asked Questions

What is bank verification software for funders?

Bank verification software for funders is technology that confirms the authenticity of an applicant's banking information during the underwriting process. Rather than trusting static bank statements that can be manipulated, modern verification software captures live evidence of banking activity. Solutions like Exact Balance use asynchronous screen recordings where applicants record their banking portal sessions, providing underwriters with video-level proof of real transactions, account balances, and account ownership.

How do platform lenders like Square achieve such low loss rates?

Platform lenders achieve low loss rates by underwriting on first-party transaction data that flows through their own payment processing infrastructure. Square sees every card swipe, every deposit, and every refund for its merchants in real time. This proprietary data eliminates the need to rely on external documents and gives their AI models extremely accurate inputs. Independent MCA funders can close part of this gap by using session-based bank verification that captures live financial data directly from the applicant's banking portal.

Can asynchronous bank verification scale with rapid MCA portfolio growth?

Yes. Asynchronous verification removes the scheduling bottleneck that limits live call-based verification. Applicants record their banking sessions at any time, from any timezone, without coordinating with an underwriter. Underwriters then review recordings in batches during focused work blocks, processing verifications two to three times faster than live calls. This model scales linearly with volume increases without requiring proportional staff increases, making it ideal for funders experiencing rapid growth.

How does AI detect fraud in bank verification recordings?

AI fraud detection in bank verification recordings works across multiple layers. Domain validation confirms the banking portal URL is legitimate. Visual consistency analysis compares page layouts and fonts against known bank templates. Behavioral analysis evaluates mouse movement patterns to detect scripted or automated sessions. Timestamp correlation ensures the recorded session matches the claimed verification period. These techniques combined make it extremely difficult for fraudsters to fabricate convincing bank verification recordings.

Conclusion

The competitive landscape for MCA funding is splitting into two tiers. Platform lenders with proprietary transaction data are posting record originations and historically low loss rates. Independent funders still relying on static bank statements and manual verification calls are losing ground on both speed and accuracy. The path forward is clear: bank verification software must evolve from document review to session-based, AI-enhanced evidence capture.

Exact Balance was built for exactly this transition. Our asynchronous screen recording platform gives independent funders verified visual evidence of live banking sessions, complete audit trails for compliance, and a workflow that scales without adding headcount. Visit exactbalance.ca to see how async verification fits into your underwriting workflow.