OSINT Research Hub › OSINT Tradecraft
The OSINT Research Method: From Question to Published Finding
A six-step, ethical workflow for open-source research: scoping, planning, collecting, verifying, analyzing and publishing findings you can defend.
Start with a research question, not a target
Good open-source intelligence begins with a question that serves the public interest: Is this casualty figure credible? Which companies supply this contract? Is this video really from that city? Write the question down before you search anything. It keeps the work focused and it protects you from drifting into collecting information you have no reason to hold.
Our publishing rule is simple: we research events, places, organizations, public institutions and public figures acting in their public roles. We do not profile private individuals, and we do not take requests to investigate a person. Research done for a paying client about a person’s character or actions is a licensed activity in New Brunswick; see OSINT, Ethics and the Law in Canada.
Plan before you collect
A short plan saves hours. List what you need to know (your intelligence requirements), the sources most likely to answer each one, how long you will spend, and what would make you stop. Decide in advance where notes, captures and files will live, and name them consistently so another researcher could follow your trail.
- Requirements: three to five sub-questions that together answer the main question.
- Source map: official records, reputable media, academic work, satellite and map data, social media, archived pages.
- Limits: time box, data you will not collect, and people you will not research.
Collect and preserve
Open sources change and disappear, so preservation is part of collection. Save a copy of every page you rely on with the Internet Archive’s Save Page Now or archive.today, take a full-page screenshot that shows the URL and the time, and store original files untouched. Recording a SHA-256 hash of each downloaded file lets you prove later that it has not been altered.
Prefer primary sources: the court decision rather than the article about it, the corporate filing rather than the press release, the original video rather than a re-upload. Note where and when you found each item.
Verify everything that matters
Verification is what separates intelligence from rumour. For each key claim, look for at least two independent sources that do not simply repeat each other. Check provenance: who first published this, when, and how could they know? For images and video, run reverse image searches (Google Lens, Bing, Yandex, TinEye) to find earlier copies, then geolocate and date the scene; see GEOINT Basics.
Many analysts grade each source with the Admiralty (NATO) system: reliability of the source from A (completely reliable) to F (cannot be judged), and credibility of the information from 1 (confirmed by other sources) to 6 (cannot be judged). Writing “B2” next to a claim forces you to think about both.
Analyze with structure
Analysts are vulnerable to the same biases as everyone else: seeing what they expect, stopping at the first explanation, trusting sources that agree with them. Structured analytic techniques push back. A key assumptions check lists what you are taking for granted. Analysis of competing hypotheses sets every plausible explanation against the evidence and looks for what disproves each one, not what supports it.
State conclusions with calibrated language (“almost certainly”, “likely”, “roughly even chance”, “unlikely”) and a confidence level, so readers know how much weight each judgment can bear.
Publish, cite and correct
A finished report leads with key judgments, then shows the evidence: every claim linked to its source or archived copy, with images annotated. Say clearly what you do not know. Disclose any use of AI tools. Blur or crop bystanders and anyone who is not the subject of public-interest reporting.
Publishing is not the end. Invite corrections, log them openly, and update the report when new evidence appears. A visible corrections record builds more trust than a report that claims to be perfect.