Key takeaways
- Evaluate the particular claim against relevant knowledge and dates rather than applying one blanket judgement to a person.
- Test independence, motives and alternative explanations using evidence that can both support and contradict the account.
- A useful validation note explains the basis, limits and effect of uncertainty on the client's decision.
Validate the claim, not an all-purpose reputation label
A person can be well informed about one process and mistaken about another. Source validation should therefore begin with the statement being used in the assessment. Identify what the person says happened, the relevant period and how the claim relates to the client's question.
Separate several issues: whether the person is relevant, how they could know the matter, whether the account is internally clear and what independent material supports it. Knowing the person exists or held a particular role does not establish the substance of every account attributed to them.
The public US Intelligence Community Directive 203 discusses source quality, access, motivation and uncertainty in government analysis. These are useful ideas for asking about analytical reasoning, not private-agency accreditation or a government requirement applying to every commercial enquiry.
A client should receive a qualified validation explanation tied to the material claims. An unexplained label such as “trusted source” should not carry a broad allegation when the report gives no reason the person could know that particular event.
Check proximity, scope and the age of knowledge
Proximity concerns the source's basis of knowledge. Distinguish direct participation or observation from an account heard from another person, general professional context and personal interpretation. All can generate useful questions, but only some can support a specific factual claim without further evidence.
Ask about time and scope. A person may have known one project, office or commercial relationship during a defined period. That does not establish knowledge of the whole business or its present management. Where the source describes a change, the report should state whether they observed it or inferred it from later information.
The ICO's accuracy guidance distinguishes records of opinion from factual information and explains the importance of source and status. Preserve those distinctions when using an account about a person; a historical opinion should not become an unqualified present-day fact.
Our HUMINT sources guide explains how different perspectives may contribute. The validation task is to assess their relevance to this claim, including what they could not reasonably establish.
Look behind agreement to the origin of the information
Corroboration depends on more than the number of people agreeing. Two sources may have attended separate events, observed different parts of the same event or repeated one colleague's story. These relationships change what their agreement establishes.
| Evidence pattern | What it may add | Question that remains |
|---|---|---|
| Two separate direct observations | Potential independent support for defined facts | Are identities, dates and the claimed event aligned? |
| Several retellings of one account | Knowledge that the account circulated | Is the original claim corroborated? |
| An account matching a narrow record | Support for the fact the record actually covers | Does it support the broader interpretation? |
| One account contradicted by a later record | A reason to review the earlier conclusion | Is this a change over time or a genuine conflict? |
The report should explain known or possible dependence. Where the origin cannot be established, describe it as uncertain rather than counting each repetition as another confirmation. A large source count can otherwise disguise a narrow evidence base.
Consider motives and competing explanations proportionately
Relevant incentives, interests or disputes can affect how an account is framed. They are factors to consider, not automatic proof that a person is truthful or dishonest. Avoid using a source's commercial position as a substitute for testing the statement.
Ask what evidence could show the account to be mistaken. A claimed current relationship may have ended. A delayed payment may have a documented contractual explanation. Similar terminology may describe different entities or events. The provider should preserve these alternatives and the records that support them.
A combined OSINT and HUMINT assignment can compare accounts with relevant available documents. The assessment should state which part of the account each document supports and where the relationship remains inferential. A shared corporate address, for example, may establish only a common address.
If material conflicts remain, the useful result may be an unresolved finding with a precise next question. Validation is not successful merely because the provider reaches a definite answer. It is useful when the client understands the evidence weight and the consequences of the remaining gap.
Apply the checks to a clearly fictional example
Fictional illustration: simulated Source A says invented Broker B controlled a supplier decision during a past project. Simulated Source C repeats the account but did not participate. A fictional record shows that B introduced one meeting, while another simulated record identifies a different decision-maker. No real source, company or completed investigation is described.
The validation note should distinguish the introduction from decision control. A's proximity may support knowledge of the meeting, while C adds no established independent observation. The contrary record materially limits the stronger control claim.
A qualified conclusion would say that an introduction is supported within the fictional evidence set, but control over the decision is not established. The next question might concern the relevant approval record. It would be misleading to present two agreeing accounts as proof of control or to omit the document naming another decision-maker.
That approach helps the client decide what to request next without treating an allegation as a finding simply because it arrived through a human account.
Explain confidence and preserve the review trail
The PHIA uncertainty framework distinguishes likelihood from confidence in the foundations of an assessment. For a commercial deliverable, ask for plain explanations of the actual evidence strengths and limits rather than unexplained precision.
Use this review checklist:
- The specific claim, event and period are identified.
- The source's relevant proximity and limits are explained.
- Direct knowledge and reported information remain separate.
- Possible common origins and relevant interests are considered.
- Documents support only the propositions they actually establish.
- Contradictions and alternative explanations remain visible.
- The note states what could change the assessment.
The HUMINT report guide shows how those qualifications should survive into summaries and annexes. If an account changes, record the change and its effect on earlier findings.
When comparing HUMINT services, ask the proposed provider to explain how it would qualify one important claim in your brief. The useful answer describes evidence and limits, rather than guaranteeing truth or relying on the provider's source network as its own proof.