# How OSINT Works

> Universal OSINT. Content reviewed 2026-10-16.
> Canonical page: https://universalosint.org/how-osint-works/

Responsible OSINT follows a repeatable process: define a question, find relevant public sources, check the evidence, analyze it in context, and communicate an answer with its limitations. Tools can assist this process, but they cannot replace judgment.

## 1. Ask a bounded question

Choose a question with a clear subject, place, and period. "What is happening to education?" is too broad. "How did the published secondary-school enrollment indicator change for a selected country over a specified period?" is a more workable starting point.

Decide what would count as an answer and what the research will not cover. Avoid collecting personal information when aggregate data will answer the question.

## 2. Locate the original sources

Start with the institution responsible for the record or dataset. Read its documentation before extracting numbers. Record the publication title, publisher, URL, reference period, retrieval date, indicator identifier, and relevant access or license terms.

For a statistics exercise, begin at [World Bank Open Data](https://data.worldbank.org/) or [the UN SDG data platform](https://unstats.un.org/sdgs/dataportal). These are starting points, not interchangeable collections. The same topic may be measured differently by different indicators.

## 3. Verify before combining

- **Provenance:** Is this the publisher's own page or a copy? Can you identify the original data provider?
- **Time:** Does the publication date differ from the period measured? Has the series been revised?
- **Meaning:** Are you comparing the same definition, population, geographic boundaries, and unit?
- **Independence:** Do two websites provide separate evidence, or do both repeat the same source?
- **Completeness:** Are observations missing, estimated, or based on a sample? What caveats accompany them?

## 4. Analyze within the evidence

Use only comparable observations. Keep missing values distinct from zero. If a definition changes midway through a series, explain the break instead of presenting a smooth trend. Consider plausible alternative explanations before attributing a change to a specific policy or event.

Separate observation ("the published value increased"), interpretation ("this may reflect improved access"), and a causal claim ("this policy caused the increase"). The last claim needs evidence beyond a simple before-and-after comparison.

## A beginner exercise: interpret an indicator

1. Choose one country and an education indicator from World Bank Open Data.
2. Read the indicator definition, unit, original provider, and available years.
3. Select two comparable years with reported values; if suitable observations are unavailable, state that the comparison cannot be made.
4. Write a short description of the change, cite the source and retrieval date, and explain at least one limitation.
5. Ask a second reader to reproduce the comparison from your notes.

This is a proposed learning exercise, not a reported finding. No figures or conclusions are asserted here. Check the live source rather than inventing values to complete the exercise.

## 5. Share an honest answer

Describe the question, method, evidence, finding, and limitations. Link to original publications and distinguish your interpretation from the publisher's conclusions. Preserve only the materials needed for reproducibility, within applicable rights and data-protection obligations.

Correct substantive mistakes transparently. If the evidence does not support an answer, "we do not know" is a valid outcome. Do not use an AI-generated summary as independent confirmation; check its claims against the original source.

**Keep the work proportionate.** This introduction is for low-risk educational research using public institutional sources. It is not a guide to tracking individuals, bypassing restrictions, or conducting high-risk investigations.

## Official sources and further reading

- [World Bank Open Data](https://data.worldbank.org/). Source for the suggested indicator-reading exercise.
- [UN Statistics Division: Global SDG Indicators Data Platform](https://unstats.un.org/sdgs/dataportal). Indicators with methodological context.
- [OHCHR and UC Berkeley: Berkeley Protocol](https://www.ohchr.org/en/publications/policy-and-methodological-publications/berkeley-protocol-digital-open-source). A detailed specialist reference on gathering, analyzing, and preserving digital open-source information.

## Continue learning

Use the [verification checklist](https://universalosint.org/verification/) and [data-reading guide](https://universalosint.org/understanding-data/). Read [the ethical boundaries](https://universalosint.org/ethics-and-responsibility/) before beginning your own project.

Original educational text: [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). Third-party sources retain their own terms.