What does an automated compliance check need from your case files?

On first calls with mortgage networks, lenders and brokers, the document question comes up early. A distribution director at a UK mortgage network asked: "Is there any requirement for standardization of documents or naming convention of the documents or anything?"

The worry is reasonable. A network with hundreds of advisers has hundreds of filing habits. One adviser scans payslips, another photographs them on a phone, a third forwards them by email. If a check only works once every file follows one standard, someone has to run that standardisation project first, and nobody budgeted for it.

An API is one route, not a requirement. The usual options:

• Export and upload. Someone exports the case from the CRM, origination platform or a shared folder and drops it into the tool. No integration is needed.
• A CRM adapter. The case goes to the check when it is submitted, and the result comes back to the case record.
• A connector for a case system with no ready-made adapter.
• An API for firms whose developers want to send cases and receive results directly.

The same check should run whichever way the case arrives. We cover the routes in more detail in our guide to mortgage CRM integration for compliance checks.

A check needs the documents your file reviewers already use. In a mortgage case that usually means:

• the fact find and the suitability letter
• bank statements and payslips
• proof of identity
• declaration forms
• call transcripts and emails, where they record what the customer said

Format matters as much as the list. Many case packs hold scanned pages and phone photos of documents, some of them poor, alongside native PDFs. A check that only reads clean PDFs pushes the clean-up back onto your team, which is the project the director wanted to avoid.

File reviewers do not grade documents in isolation. They check that the income in the fact find matches the payslips, that the recommendation in the suitability letter fits the needs in the fact find, and that the evidence supports both.

So ask whether a tool reads the case as a whole and cross-references the documents, or extracts each document on its own. Extraction alone tells you what a payslip says. It does not tell you whether the case holds together.

The FCA's Consumer Duty guidance (FG22/5) says firms "will need to identify relevant sources of data to enable them to assess whether the outcomes that their customers are experiencing are consistent with their obligations under the Duty". It also accepts that "file reviews will be based on a risk-based sample".

Case files are one such source, but only if they can be read consistently. When every case goes through the same check, whatever its format, the results show which advisers miss which documents, how often and on what kind of case. A sample built around the files that are easiest to review cannot show that.

1. Take a handful of recent cases from different advisers and export them the way you would for a check.
2. Note what each export holds, and what sits outside it, such as emails or call notes kept elsewhere.
3. Note the formats: native PDFs, scans, phone photos, transcripts.
4. Decide who exports cases today and at which stage: before submission, after submission or at file review.
5. Take those notes to any vendor conversation, so the questions below get specific answers.

1. Does it need a naming convention or folder structure, or does it work out what each document is from its content?
2. Which formats does it read: scans, phone photos, call transcripts, emails?
3. Does it read the case as a whole and cross-reference the documents?
4. What does the result say when a document is missing or a page cannot be read?
5. Does the same check run whether the case arrives by upload, API or CRM adapter?

Curvestone reads any document expected in a case pack, including scanned and badly photographed documents, call transcripts and emails. It reads the case as a whole, cross-referencing the fact find, the suitability letter and the evidence. Drag and drop needs nothing installed and no integration project, and the same check runs whether a case arrives that way, through the API or through a CRM adapter such as Finova or OMS. To see how it reads a case, book a call or a demo.