What a waterfall actually does
A lending waterfall routes an applicant through an ordered set of financing options, often after an initial lender declines or does not produce an acceptable offer. It can expand access by matching different underwriting appetites. It also changes which applicants each lender sees, how long the process takes and which terms the consumer ultimately encounters.
ServiceTitan’s product documentation describes a second-look workflow that can continue after a first-look decline and evaluate additional financing options. [1][2] That establishes a documented commercial use case, not an independently measured approval or consumer-benefit claim. The analysis here concerns the mechanism and controls, rather than recommending a particular provider or reproducing its advertised conversion figures.
Selection happens before the second lender decides
The second lender does not usually receive a random sample of all applicants. It receives people filtered by the first lender’s policy, customer choices and the platform’s routing rules. A change in first-look cutoffs can alter second-look performance even if the second lender changes nothing. Pricing or loss estimates based on an old referral population may then become unreliable.
Routing also creates missing outcomes. If a customer abandons after the first decline, the platform cannot assume a second lender would have approved or funded the transaction. If the customer receives an offer but rejects its terms, counting that as a successful conversion overstates commercial results. Keep eligibility, application, approval, acceptance, funding and completed purchase as separate stages.
An effective data contract preserves the routing version, lender order, timestamps, requested amount, offers and customer choices. A lender should know whether it is evaluating a direct applicant or a referred applicant with a prior outcome, subject to lawful information sharing. The platform should not silently change the applicant pool while reporting performance as if the channel were unchanged.
A hypothetical approval calculation
Start with 1,000 completed first-look applications. Assume Lender A approves 600. Of the 400 declined, 300 choose to proceed to Lender B, which approves 150. Unique applicants with at least one approval total 750, or 75% of the initial population. Lender B’s approval rate is 50% of its 300 applications, not 15% or 37.5% unless those alternative denominators are explicitly labeled.
Now assume 500 of A’s approved applicants and 90 of B’s approved applicants actually fund. The combined funding rate is 59%. If 20 funded purchases are subsequently canceled, completed transactions are a further distinct measure. These assumed figures illustrate why an approval claim cannot be treated as revenue or customer value without following the rest of the funnel.
The incremental funding contribution of B is 90 accounts in this example, but causal incrementality remains uncertain. Some customers might have obtained financing elsewhere or paid another way. A controlled rollout or carefully matched comparison can help assess the true lift, provided the design is lawful and does not withhold required treatment or notices.
Preserve the consumer’s decision
An ordered lender list may optimize merchant fees, approval probability, expected funding, customer cost or platform compensation. These objectives can conflict. A route producing the highest approval rate may lead to a more expensive offer or a different legal product. Make the customer’s available choices and material terms understandable before commitment.
Do not describe a lease, a revolving line and a closed-end installment loan as interchangeable merely because each enables a purchase. Compare down payment, total payments, ownership, cancellation, deferred-interest conditions and other material features where applicable. An apparently lower monthly payment may result from a longer term rather than a lower cost.
Regulation B §1002.9 contains adverse-action notification requirements and provisions addressing applications submitted through third parties to multiple creditors. [3] The actual structure matters. A platform should assign responsibility for notices and preserve each creditor’s decision rather than assuming one generic platform message always satisfies every obligation. This article does not determine coverage for every possible routing arrangement.
Economics at the merchant and lender
For a merchant, incremental gross profit from a completed sale must cover financing fees, platform cost, returns and service expense. A waterfall can increase booked sales while reducing net margin if the marginal financing option is expensive or associated with higher cancellation. Measure outcomes by financed product and lender route, not just the total number of offers displayed.
For a lender, a rejected-first-look population can be attractive if its own information or product design addresses risks the first lender does not serve. It can also concentrate adverse selection. The OCC’s retail-lending handbook provides a supervisory foundation for disciplined underwriting and portfolio monitoring. [4] A referral arrangement does not replace the receiving lender’s credit judgment.
Recommended tests compare approved and funded populations over time, controlling for channel and vintage where feasible. Watch first-payment defaults, fraud, merchant disputes, prepayment and repeat borrowing. A change in first-look lender policy should be treated as a potential model or strategy input change for downstream lenders.
Failure modes and useful evidence
Operational failures include duplicate applications, stale offers, unexplained credit inquiries, notices that identify the wrong creditor and routing loops after a service outage. Use unique application identifiers and idempotent submissions, with clear expiration and retry rules. A fallback path should not send sensitive information to a new party without the applicable authority and disclosures.
Evidence of success would include incremental completed purchases, sustainable repayment, transparent offer comparisons and lower abandonment without worse customer outcomes. Vendor approval statistics alone cannot establish those results. The most useful waterfall is one whose selection effects, economics and legal responsibilities remain visible as lender order and market conditions change.
Sources
- ServiceTitan, second-look financing with TURNS overview; product documentation reviewed September 27, 2026SourceBack to text: ↑
- ServiceTitan, single-application waterfall financing documentation; reviewed September 27, 2026SourceBack to text: ↑
- CFPB, Regulation B §1002.9, including multiple-creditor notifications; reviewed September 27, 2026Official textBack to text: ↑
- OCC, Retail Lending handbook, October 2021; reviewed September 27, 2026Official source · PDFBack to text: ↑