Uncovering hidden reports and untold stories
Uncovering hidden reports and untold stories
Seven Common Mistakes in Corporate Records Investigations

Seven Common Mistakes in Corporate Records Investigations

On October 15, 2026

Corporate records have a reputation for being the most reliable evidence in the room. They are dated, stamped, filed, and in many jurisdictions publicly searchable. That authority is exactly what makes them treacherous. A file that looks complete invites you to stop asking questions, and most of the failures below happen at that precise moment — not because the investigator was lazy, but because the paperwork looked settled.

What follows are the mistakes that surface again and again when a records investigation is reviewed by a lawyer, a regulator, or an opposing party who has done their homework.

Mistake One: Believing the Registry Over the Company

A company registry records what was filed, on the day it was filed. It is not a live picture of a business. Filings arrive late, changes are never submitted, and the address on file is often the office of a formation agent rather than anywhere a director has ever worked.

A simple example. You pull an extract in December and see three directors. The company's own board minutes from July reference a resignation. Neither document is wrong. One is simply older than the other, and you will not know which without asking for the filing acknowledgement, the date of the last update, and the correspondence that triggered it.

A few habits worth adopting:

  • Record the exact date and time you obtained every extract, and say so in your notes.
  • Order a historical extract where the registry offers one, and compare it against the current version.
  • Ask the company for proof of its most recent filing, not just a description of it.
  • Treat any field that changed without a corresponding filing as a question, not a finding.

Mistake Two: Stopping at the Parent Company

Group charts are marketing documents. They are selective, frequently years out of date, and tend to omit the entities set up for a single contract, a single property, or a single purpose nobody wants to explain. Investigators who build their scope from that chart will miss the subsidiaries that matter.

Where the missing entities hide

Look at the shareholders section of every entity you have, then search the names you find there as if they were companies in their own right. Ownership often runs through a holding company registered in another jurisdiction, a trust, or a nominee arrangement. Names drift too: "Northbridge Trading Ltd" and "Northbridge Trading Company Limited" may be the same business or two unrelated ones, and only the registration number settles it.

Search abbreviations, former names, transliterations, and the personal names of directors. Then run the exercise in reverse. Some entities claim a parent they do not have, and a claimed relationship that appears nowhere in the filings is itself worth a paragraph in your report.

Mistake Three: Accepting a Document Trail With Holes in It

Most disputes about records are not about what a document says. They are about where it came from. A signed resolution arrives as a scan with no accompanying email. Minutes are undated. A contract is produced in three versions and nobody can say which one was executed. Each gap is small; together they can make a conclusion indefensible.

Ask for the native file, then ask again

Scans lose metadata, and metadata is often the most useful part of the file: creation dates, authors, revision history, the trail of who sent what to whom. Email headers do similar work. Where you can obtain original files lawfully, keep them alongside the PDFs and preserve both without editing. If a signature page has circulated separately from the resolution it belongs to, note it. On its own it proves nothing, but it changes the questions you ask.

Paperwork does not prove an event. It proves that somebody wrote something down.

Mistake Four: Confusing What the Records Say With What Happened

Board minutes are usually approved at the next meeting, sometimes months later, and are occasionally rewritten to reflect what people wish had occurred. Undated documents get dated by whoever assembles the bundle. A file can be internally consistent and still describe a meeting that never took place.

A useful discipline is to separate three things in every note: what the document states, what it implies, and what has been independently confirmed. Reports collapse all three into one confident sentence, and that sentence is where the trouble starts. Write "the resolution records that the transfer was approved on 4 March" rather than "the transfer was approved on 4 March", and the distinction stays visible for whoever reads the file next.

Mistake Five: Reaching the Conclusion Before the Boring Work Is Done

The first document that makes sense of a messy file is hard to let go of. It explains the payments, the resignation, the missing subsidiary, all at once. From that point, every new record gets read as confirmation, and anything that contradicts the theory is quietly deprioritised.

Guard against it deliberately:

  1. Write your leading hypothesis down, in one sentence, before you finish reviewing the file.
  2. Write a rival explanation that fits the same documents, and list what would distinguish the two.
  3. Build a dated timeline of filings, payments and correspondence. Gaps are findings too.
  4. Set a rule: no conclusion until the last three years of filings are reviewed and written questions have been answered or refused.

Refusals are data. A company that will not answer a straightforward question about a subsidiary has told you something, even if it has told you nothing about the subsidiary.

Mistake Six: Leaving Out the People Who Made the Records

Registers, minutes and filings are produced by company secretaries, in-house counsel, finance staff and outside advisers. Those people are usually more informative than the documents they created, and easier to approach than investigators assume. Former employees in particular often know where the file is thin.

Approach them as people, not as witnesses. Explain what you are looking at, ask open questions, and do not open with an accusation. Keep a same-day note of who said what, on what date, and whether they were speaking from memory or from documents. Where a source needs protection, agree the terms before the conversation, not after.

One caution worth stating plainly: interviews can be legally sensitive, and obligations around confidentiality, employment and personal data differ by jurisdiction. If a conversation could end up in a formal proceeding, take advice before you have it rather than afterwards.

Mistake Seven: Keeping a Sloppy File About Your Own Investigation

The weakest document in many investigations is the investigator's own working file. Requests go out by phone. Documents arrive in a chat message and get saved as "final_v3". Notes are rewritten at the end of the week from memory. Then someone asks how a particular conclusion was reached, and the answer cannot be reconstructed.

Keep a simple register: what was requested, from whom, on what date, what was received, and where it now sits. Save files as received, without renaming them beyond an identifier, and keep an untouched copy. Write notes the same day. Where you rely on a judgement, say so, and say what would change your mind.

A Short Checklist Before You Sign Off

None of this requires unusual resources. It requires a willingness to look again at the things that seem finished.

  • Current extract, historical extract, and proof of the most recent filing for every entity.
  • A group structure built from filings, not from the company's own chart.
  • Native files and email headers wherever they can be obtained lawfully.
  • A dated timeline with the gaps left visible rather than smoothed over.
  • A named source for every document, and a note of how it arrived.
  • Written questions sent, and answers or refusals recorded.

Finally, keep your language proportional to your evidence. "The records do not show" is a finding. "It did not happen" is a claim, and it needs more than a missing page behind it. Where an investigation is likely to end up in a dispute, in a regulator's hands, or in court, have a lawyer review the file before it travels.

Photo: cottonbro studio / Pexels

Comparing Complaint Databases: How to Spot Patterns Across Local Agencies

A practical guide to comparing local complaint databases: where records live, how to standardize them, and how to surface repeat problems, delayed...

From Tip to Story: A Beginner's Guide to Structuring an Investigation

A tip is not a story. Here is the practical path from an anonymous message to a documented investigation: timelines, verification, source care and...