The exact evidence an auditor asks for in week one — SOC 2 access reviews, AML transaction-monitoring fields, Reg BI audit data — checked off online or exported before the engagement starts.
An SEC examiner walks each case asking three questions: did you have the data to evaluate suitability, did your process use it, and can you show the audit trail? This is the field list that gets you to yes on all three.
Onboarding workflows built for the modal client — W-2, US-citizen, one state — quietly reject H-1B holders, ITIN filers, and dual residents. Losing them isn't a compliance requirement, it's a competitive miss.
First-time SOC 2 audits stall on the same predictable gaps — a missing access review, a de-provisioning ticket that's three weeks late, a sub-processor with no report on file. This checklist is the evidence list your Type II auditor actually walks through in week one.
A corpus can pass a smoke test and still mis-train a model because nobody checked age against income, or paired a nuclear engineer with a hospitality job. This is the fidelity gate that catches it before production does.
A monitoring system fed only amount, date, and counterparty catches obvious structuring — and misses cross-account aggregation, layered transfers, and PEP-adjacent activity. Here's the field list that closes the gap.
Stale AUM, mis-categorized client counts, an undisclosed disciplinary update, a brochure that no longer matches your marketing — the deficiencies the SEC cites most, and the data prep that prevents each.
A documented data inventory, a risk assessment mapped to controls, and a notification clock that starts at discovery, not containment — the FTC's Safeguards Rule requires all three, no exceptions.
Wealth-tech launches rarely go sideways on the engineering — they fail on a missing compliance review, a disclosure that wasn't refreshed, or a CS team that never got briefed on the failure modes.
Feature worked great in testing, then a real customer did something the corpus never modeled — the fix is quick; the trust hit and rollback aren't. Those edge cases are enumerable; this is the audit list.
Robo-advisors must onboard fast and still produce a best-interest recommendation — the SEC's 2021 sweep found recurring gaps in profiling depth, allocation mismatch, and a vanishing audit trail after onboarding.
Form CRS delivered late, or a conflict disclosed in the brochure but never surfaced at recommendation — the 2024 SEC sweep's simplest findings all trace back to one missing timestamp or acknowledgment.
A vendor's benchmark slide rarely says who wrote the ground truth, whether a held-out set actually exists, what retrieval budget was used, or which model created the answers. This checklist is the five things to ask before trusting the number.