Who this is for. A plain-language map of Sooner's data collection — built to be skimmed, not read like code.
Debt providersDiligence on what we collect, what is verified, and where each underwriting input comes from.
Sooner teamThe single source of truth for what the flow asks and why.
New employeesOnboarding to the product without reading the codebase.

The flow, end to end

Five stages. A borrower is qualified fast at Form 1, then completes a detailed financial profile, an optional co-applicant profile, a property simulator, and finally uploads documents that verify what was declared. Nothing the borrower types is trusted on its own — the underwriting score firms up only once documents are in.

How to read every table

Collected as
Borrower-declared  Typed or selected by the borrower in the flow.
Prefilled from Form 1  Carried over from lead-gen; borrower confirms or edits, never re-types.
Document-verified  Substantiated by an uploaded document at Form 3.
Derived  Computed by the flow, not asked directly.
Requirement & what it feeds
Required  Always asked; the borrower cannot proceed without it.
Conditional  Only asked when a branch condition is met (shown inline).
Optional  Borrower may skip.
G9 a hard eligibility gate   Cat A a scoring category. Hover any tag for its meaning.

The underwriting framework these answers feed

Two mechanisms decide an application. Gates (G1–G14) are hard pass/fail checks — fail one and the application is declined or sent to committee. The Sooner Credit Score (SCS) sums ten weighted categories (A–J) to 100; 70+ approves, 55–69 goes to committee, below 55 declines. Pricing is a separate, service-driven fee ladder.

Scoring categories & weights
Service fee ladder

Fee is driven by which services the borrower selects — not by their score. Closing Fee Financing (CFF) is required for any offer.

Decision thresholds — SCS ≥ 70 approve · 55–69 committee · <55 decline. AECB ≥ 550 or a thin-file path is required to score at all.

Collection vs Verification matrix

The centerpiece. Every substantive data point, one row each: the question as the borrower sees it, the possible and accepted answers (the option set plus the validation rule, and the underwriting band that actually qualifies), how it is collected, which document verifies it, whether it is required, and the gate or scoring category it feeds. A tick marks the range that qualifies, as distinct from the range the form merely accepts. Search or filter to narrow.

Stage
Collected
Question (as the borrower sees it) Possible / accepted answers Collected as Verified by Requirement Feeds

Questions by form

The full inventory, stage by stage, with every option spelled out and the exact branch condition for each conditional question. This is the verbatim question set — useful when a reviewer needs to see every enum value and the precise logic that shows or hides a field.

Documents & verification

The verification side. Each document tile, when it is required, the fields it confirms or extracts, the OCR vendor path, the confidence target, and what happens when confidence is low. Borrower-declared figures are cross-checked against these; conflicts route to a human reviewer.

Document Required for Acceptable upload Fields verified / extracted Confirms (gate / category) OCR vendor path Target conf. If confidence is low

Upload mechanics worth noting

Why we ask — gates & scoring

The reference behind the "Feeds" column. Gates are hard pass/fail checks; a failed gate declines or routes to committee regardless of score. Scoring categories contribute a weighted share of the 100-point Sooner Credit Score.

Eligibility gates (G1–G14)

Sooner Credit Score categories (A–J)

Weights sum to 100. The thin-file path (no AECB history) shifts Category F to 0% and Category G to 15%.

Code Category Weight What "excellent" looks like Primary inputs

Service fee ladder

BundleFeeServices included

Honesty & known gaps

Where the live flow does not yet match the eventual underwriting engine, or where collection runs ahead of scoring. Stated plainly so diligence does not have to discover it. None of these affect what is collected — they affect how (or whether) an input is yet used.