Key takeaways
- Define the missing human perspective before comparing agencies.
- Verify the proposed team, source relevance, validation and method boundaries.
- Agree reporting, confidentiality and scope changes without demanding a preferred conclusion.
Define the human-source question
Choosing a HUMINT agency starts with explaining what a person with relevant experience could add. You may need accounts of a management decision, a commercial relationship or conduct that is poorly documented. State the question and why existing records leave it unanswered. A request to discover everything about a company does not define a useful human-source assignment.
In a hypothetical corporate matter, a board wants to understand whether managers discouraged staff from reporting concerns. Records may show formal policies but not how employees experienced them. The commission should identify the relevant period, organisational roles and reported behaviour, without assuming that a particular allegation is true.
Our commercial HUMINT services comparison can help identify teams to approach, but confirm the actual proposed work. A provider’s general intelligence description does not establish that it will conduct suitable human-source enquiries for your case.
Use the corporate HUMINT guide to distinguish a people-based workstream from a broad corporate report. Ask whether the missing answer is realistically available through relevant accounts, which documents could test them and what would remain unresolved if appropriate contributors are unavailable.
Verify the team and the relevant capability
Ask who will lead the assignment, conduct enquiries and review the analysis. Relevant language and sector knowledge should relate to the actual source population and subject matter. Clarify whether other organisations or subcontractors will participate and how their work is supervised.
Look for evidence of a specific collection method rather than a general promise of access. Ask who conducts the interviews, how direct knowledge is assessed and which records can support the account. A useful proposal explains how source reporting becomes a qualified finding. Compare that explanation against your question, not against another company’s marketing example or an assumed standard package.
Discuss a hypothetical version of your problem and ask how the proposed team would assess relevance, distinguish direct experience from repetition and handle conflicting accounts. The answer should explain method and limitations, rather than rely on impressive biographies or a promise to find a particular number of people.
Confirm conflicts involving the client, subjects and relevant intermediaries. Establish who receives material findings and who approves changes. A source enquiry concerning management should have a reporting route that does not silently depend on the managers whose conduct is being examined.
Compare source assessment and validation
A useful proposal explains how source accounts will be evaluated. Ask how the team records a contributor’s relevant experience, recency, basis of knowledge and possible interests. Several accounts may repeat one story; a direct account can still be mistaken or incomplete. The proposed validation method matters more than an unexplained contact total.
| Area | Ask the team | Expected explanation |
|---|---|---|
| Relevance | Why could this source know? | Role, period and connection to the question |
| Knowledge | What was directly experienced? | First-hand account separated from hearsay |
| Independence | Do accounts share an origin? | Common sources and repeated stories identified |
| Corroboration | What else can test the account? | Relevant records and alternative explanations |
| Limits | What cannot be established? | Uncertainty visible in the findings |
The source validation guide develops these distinctions. Ask for them in the deliverable rather than assuming the word HUMINT guarantees verified information. An informative interview can provide a lead without establishing the underlying event or another person’s intent.
Agree method and confidentiality boundaries
Define permitted contact, necessary personal information, handling responsibilities and any method exclusions before outreach. The team should explain the proposed approach at an appropriate level without exposing contributors unnecessarily. A commercial mandate is not authority to obtain protected records or breach another person’s access restrictions.
The ICO’s announcement of the investigators’ data-protection code identifies issues including lawful processing and controller or processor responsibilities. Turn those into engagement questions about this work: who controls the records, which parties receive them and how access and retention are agreed. Do not assume that a code announcement certifies the proposed agency or your particular assignment.
Set realistic confidentiality expectations. Ask what information the client will receive about source relevance and provenance, what identities may be restricted and whether the intended use requires another evidence process. An absolute promise that no one can ever infer an enquiry occurred is not a useful contractual substitute for explained safeguards and limits.
If public records are needed to test accounts, define that supporting work through the combined OSINT and HUMINT guide. Keep the human-source and documentary contributions distinguishable so the reader knows what each establishes.
Specify the reporting standard
Request an outline of the final report before accepting the commission. It should explain the question, enquiry scope, source relevance, material accounts, corroboration, contradictions and limitations. Agree how the client can assess provenance without unnecessary disclosure of identities or unrelated personal information.
A report should distinguish observation, reported experience and analytical judgement. The ICO’s accuracy guidance discusses recording sources and identifying opinions appropriately; the guidance currently notes that it is under review following legislative changes. For this engagement, require clear attribution and a process to address material corrections rather than presenting every account as an established fact.
In the hypothetical management-culture matter, one person may describe a specific meeting while another expresses a general view of the organisation. Both may be relevant, but their evidential reach differs. Ask the team to explain that difference and identify records or other accounts that could test a material claim.
Agree who will receive interim findings and whether a clarification discussion is included after delivery. If the intended use changes to formal proceedings or another process, obtain appropriate advice and reconsider the evidence requirements. A commercial intelligence report is not automatically a formal witness statement or expert report.
Use a final selection checklist
Choose against the agreed question and evidence standard. An agency should be able to explain why its proposed source enquiry is appropriate and what would count as completion, including the possibility of limited access or inconclusive accounts. The checklist below makes that assessment reviewable.
- Purpose: a defined question requiring human experience or perspective.
- People: named case leadership and relevant enquiry and review capability.
- Scope: subjects, period, geography and supporting records.
- Validation: relevance, first-hand knowledge, independence and corroboration.
- Boundaries: approved methods, information handling and confidentiality limits.
- Output: findings, contradictions and uncertainty in the agreed format.
- Control: checkpoints, changes and a responsible approval contact.
Ask shortlisted teams to explain one credible obstacle and how they would report it. A source who is unavailable or lacks direct knowledge should lead to a transparent limitation or an approved alternative, not a padded list of contacts.
Select the proposal that offers a credible evidence route and honest limits. The goal is a better-supported decision, not a guaranteed favourable account, a predetermined allegation or the largest possible collection of opinions.