How do you verify information?

Get back to the source, date it, cross-check, identify who's speaking: the method newsrooms use, turned into simple steps you can run yourself.

Updated on 16 July 2026

Verifying information isn't deciding whether it sounds plausible. It's establishing where it came from, when it was produced and who stands behind it. That shift is essential: plausibility is an impression, provenance is a fact you can check. Newsrooms work no differently, and the bulk of their method needs no professional tooling.

The four questions to ask

Who says so? Not which account shared it, but who asserts it in the first place. A share isn't a source; it's transport.
How do they know? A direct witness, a document, a measurement — or "someone told me"? This question alone eliminates a large share of false claims.
When? An accurate item can be misleading if it's three years old and presented as current. It's the most effective disinformation technique precisely because nothing in it is false.
Who else says it, independently? Three outlets quoting the same single source aren't three confirmations. They're one, repeated.

Getting back to the primary source

Most of what we read is second- or third-hand narration: one site reports what another site wrote about a report. At each step a detail is lost or hardened. The discipline is to go back one rung: if a piece cites a study, find the study; if it cites an official figure, find the publication carrying it; if it cites a statement, find the video or the transcript. Nine times out of ten the primary source is two clicks away, and nine times out of ten it's more nuanced than what was drawn from it.

The special case of images

The overwhelming majority of misleading images aren't doctored: they're authentic but stripped of their context or their date. A real photo of a real protest, presented as yesterday's when it's five years old, contains not one false pixel. So the useful reflex is reverse image search, which shows where the picture has already appeared and since when. That single move settles most cases.

Signals that should slow your reading down

The content triggers a strong, immediate emotion. That's rarely accidental: what's built to travel is built to make you react before you think.
No source is named, or only "experts", "a study", "sources close to the matter".
The text has no author, no date and no identifiable legal notice.
The item confirms exactly what you already thought. It's the hardest signal to apply to yourself, and the most useful.

A three-minute method

Run a search on the keywords of the fact itself, not on the article's headline. Look at who else is reporting it and since when. Find the primary source's name in the text, and open it. If the fact is real and recent, you'll have independent confirmation in under three minutes. If you find nothing, or only sites citing each other in a circle, you haven't proved it false — but you now know it isn't established yet, and that's exactly the information you needed before sharing it.

Frequently asked questions

How many sources does it take for something to be confirmed?

The traditional rule is two sources independent of each other. The important word is "independent": two people from the same department repeating the same line count as one. An original document is often worth more than three matching accounts, because it doesn't misremember.

Can you trust fact-checking sites?

They're outlets like any other: they can be wrong, and they too have an editorial line. What makes them useful isn't their verdict but their method: a good fact-check shows its sources and lets you retrace the path without it. If a verification piece asks you to take its word for it, it's no better than what it claims to correct.