Methodology

A sample you can defend, grades that map to your policy, and a report that does not need a translator.

This page is the working method behind every file-review engagement. It is not a software platform. If the method does not fit the question you have, we say so before you send files.

Organised document folders prepared for a structured review

1. Lock the universe before anyone pulls files

The sampling frame is written down: product variant, origination channel, merchant group if the question is shop-floor, approved versus declined versus withdrawn, and the date window. We exclude files only for documented reasons — test accounts, staff plans, or a product that was not live. Convenience samples of ‘typical good approvals’ are rejected.

Sample size follows the question, not a poster percentage. A disclosure check across two till scripts may need fewer files than a full application audit across marketplace and in-store channels. We say the minimum we need to speak with any confidence, and we say when the population is too thin to support a programme-level statement.

2. What each file is read for

Every selected application is scored on the same axes. Missing evidence is recorded as missing, not inferred from a later-performing account.

Capture

Identity evidence, device or staff identifiers, timestamps, and whether the flow was customer-led or merchant-assisted.

Capacity

Income method, existing instalments including other BNPL, and whether living-cost assumptions were written down.

Explanation

Versioned terms, scripts, fee and cooling-off language, and acknowledgements that can be tied to this origination.

Decision

System outcome, manual steps, overrides, and whether the reason code matches the policy pack dated that day.

3. Finding grades

Observation — a documentation habit that should change but did not decide the file.

Minor — a gap that would have been fixable at origination with the evidence then available.

Major — the recorded decision is not supported by the file, or the customer-facing explanation contradicts the internal reason.

Material — a pattern across the sample, or a single file where identity, capacity, or consent cannot be established.

Grades are tied to your written rules. We do not import another lender’s handbook and call it industry practice.

4. The report you walk into a committee with

Each file has a worksheet. The report summarises grades, concentrates findings by channel or merchant group, and lists fixes that operations can own. Remediation and re-underwriting are quoted only after this report exists. There is no dashboard login and no licence key.

If you already know the question is a full application audit, start there. If you only need the method applied to disclosure or declines, say so in the request — we will not upsell a wider universe for the sake of a thicker binder.

Send a sampling question See the engagement types