Source credibility

There is no single accepted scale. Analysts typically consider several questions. Does the source publish under a named author or editor, and does it disclose ownership and funding? Does it correct errors openly? Does it cite primary evidence? Is it independent of the subject it covers? What is its history of accuracy on this topic? For accounts on social platforms, additional signals include account age, posting pattern, follower composition and whether the account has been linked to coordinated activity.

Some organizations publish ratings of news outlets. These ratings use differing methods, and outlets sometimes dispute them, so a rating is best read as one input and its method should be cited alongside it.

A hypothetical case: a claim about a supplier's safety record appears in a local news report and in a blog with no named author. Both are counted as mentions in a media monitoring feed. The report cites inspection records and quotes a regulator. The blog cites nothing. A team that weights them equally will overstate how widely the claim is established.

Mention counts treat all sources alike unless the analyst adds a layer. A credibility view lets a team separate a claim carried by established outlets from the same claim circulating in a small cluster of low-quality sites. It affects prioritization, escalation and whether a response is worth the attention it will draw.

The same idea applies to AI answers. Assistants such as ChatGPT and Perplexity cite the sources they draw on, so the credibility of the sites they cite shapes what they say. That is why the cited domains and share of citations matter for visibility, and AI visibility is partly a question of which sources are treated as authoritative.

Many teams keep a tiered list of outlets and accounts for their topic, with notes on why each is trusted or treated with caution. The list is reviewed on a schedule and whenever a source changes ownership or has a public correction. Tiers help analysts decide quickly which mentions to escalate. They are a working aid for prioritization and not a public rating, and they should be documented so that a second analyst would reach the same result.

When an analyst rates a source, the rating should come with the reasons and the date. Notes such as "no named authors; ownership not disclosed; no corrections page" are more useful than a score alone, since they can be checked and revisited. A written rationale also lets a colleague or a client review the decision.

Credibility is not the same as agreement. A source can be credible and critical of your organization. It is also topic-specific: an outlet with strong reporting on finance may be weak on science. And it is not fixed. Sources change owners, standards and staffing, so past ratings can go stale.