The report is where careful work goes to die. Weeks of disciplined collection get compressed into eight pages written at speed, and the hedges disappear first. Then someone acts on it.
A report that survives scrutiny follows a few rules, none of them complicated.
Separate what you established from what you infer
Three categories, kept visibly apart:
- Established. Supported by admitted evidence, with references.
- Assessed. Your professional judgement, labelled as judgement, with the reasoning shown.
- Unresolved. Open questions, including the ones you ran out of time to close.
Readers do not object to uncertainty. They object to discovering it later.
Lead with the answer
The first page answers the question that was asked, in the terms it was asked. Everything else supports it. An investigative report is not a mystery novel, and burying the conclusion on page eleven guarantees it will be misread by whoever skims it.
Make every claim traceable
Each statement should point to the material behind it. If a reader cannot get from a sentence to a source, the sentence is an opinion — which is fine, as long as it is labelled as one.
This is why the report should be assembled from findings that were already reviewed rather than written from memory. See the Archive and Engramite Writer.
Say what you did not find, and where you did not look
Scope limitations belong in the report, not in a follow-up email. "Corporate records in jurisdiction X were not accessible" changes how the reader should weigh everything else.
Absence of evidence is worth stating explicitly: "no evidence was found that X" is not "X is false", and the difference has ended cases.
Write about people carefully
Name people only where necessary, describe allegations as allegations, and keep out the detail that serves curiosity rather than the question. Reports get forwarded, and a sentence written loosely about a private individual travels further than you intend. See provenance and retention.
The checks before it goes out
- Does the summary answer the question that was asked?
- Can every established claim be traced to admitted material?
- Is judgement labelled as judgement?
- Are identity claims supported by an exact link, not a shared name?
- Are scope limits stated?
- Would this read acceptably to the subject's lawyer?
That last one is the real test.
Frequently asked questions
How long should a report be? As long as the question needs. Length is not rigour.
Should raw material go in the annexes? Enough to support the findings; the full case file stays where it lives, under its retention rule.
Who should review it? Someone who did not do the collection. Fresh eyes catch the leap from "probably" to "is".
Read next: accepted and rejected evidence or what is OSINT.