01 · Intake
Name the question
We ask what you already suspect: income construction, override hygiene, group disclosure, policy drift, or a general pre-score health check. A vague “look at quality” becomes a scoped question before any file is copied.
Home / Methodology
Methodology
This is the custom working page for the practice: not a catalogue of products, but the sequence we actually follow when a lender asks us to read a sample of credit applications.
01 · Intake
We ask what you already suspect: income construction, override hygiene, group disclosure, policy drift, or a general pre-score health check. A vague “look at quality” becomes a scoped question before any file is copied.
02 · Scope note
Product, origination window, sample size, and documents required go into a short note with timeline and fee. The sample is designed for the question. It is not dressed up as a census of the book.
03 · Transfer
You send copies under the engagement letter — paper, scan, or a controlled folder. We log what arrived. Missing exhibits are recorded as missing; we do not invent them from outside the file.
04 · Reading
Each pack is read as a financial document. Stated income, commitments, purpose, related parties, and exception papers are traced to exhibits. Contradictions go on an exception log with a severity note.
Before the report is frozen, the engagement lead walks through examples with the people who own origination or second-line review. The aim is recognition: originators should see their own packs, not a generic defect taxonomy.
You receive a report that states the question, the sample, the method, the findings with examples, and the limits of the work. We will not extrapolate a thirty-file sample into a book-wide certificate. If you need a statistical design, that is a different letter.
Reports are written so they can be tabled at credit committee or shared with internal audit. They are not press releases and they are not scorecard papers.
We do not call the scoring engine. We do not interview applicants. We do not issue approve or decline recommendations. We do not opine on whether a scorecard is statistically sound. Those limits keep the report about the application file — which is the object you asked us to audit.
If the sample uncovers something that looks like misconduct rather than construction error, we stop the ordinary findings path and speak to the named counterpart. We do not run a forensic investigation unless that work is separately instructed.