The idea, in plain English
Start here.
Fact-checking starts with a specific claim. Identify who produced the evidence, open the original material and check whether it supports the exact wording you plan to publish. Dates, methods, scope and limitations matter. A polished page, a citation or agreement with your expectations is not enough by itself.
A draft says ‘Customers prefer this design.’ There is a link beside it, so the sentence looks supported. But the linked note describes a small practice survey, not every customer. This lesson shows how to move from a confident sentence to a precise, checkable statement without burying the reader in qualifications.
A few words you’ll meet.
OPEN WHAT YOU NEED ↙Claim
A statement that can be checked against evidence, such as a product specification, a statistic or a description of how a feature works.
Primary source
Original material for the question being checked, such as the actual study, specification or official product documentation. Its scope and quality still need review.
Lateral reading
Leaving a page to investigate its author, publisher or claims using other sources.
Evidence scope
The people, products, conditions, time and question the evidence actually covers.
Pull one checkable statement out of the paragraph
Underline the part that a reader could reasonably ask you to prove. A paragraph may contain several claims: a feature exists, it is available on a particular plan and it improves an outcome. Those require different evidence. Split them before searching for a source so one useful citation does not become a blanket endorsement.
Write the claim in neutral language. Words such as ‘always’, ‘everyone’, ‘best’ or ‘guaranteed’ expand what you need to establish. Ask whether the source could actually support that breadth. Keep opinion and recommendation distinguishable from a factual specification or measured result.
Match the source to the job
Check who created the material, what relevant expertise or access they have, why it exists and how it was produced. A provider’s current documentation may be the right place to check its own feature rules. A vendor’s promotional claim about superiority needs different scrutiny. An article can be useful background without being strong evidence for every assertion it contains.
Then investigate outside the page when needed. Look for information about the organization, credible coverage and the original document behind a repeated claim. Several pages repeating the same source are not several independent studies. A professional design, familiar domain ending or high search position does not settle accuracy.
- Identify the author or responsible organization.
- Check relevant expertise, method, purpose and date.
- Ask whether other coverage is independent or merely repeats the same original material.
The sample is part of the statement
Read the supporting passage and its limits
Open the original source and locate the passage, table or specification that matters. Check the date, definitions, sample and conditions. A result about a particular group or product version should not silently become a statement about everyone or every version. An older source is not automatically wrong, but it must still fit the current claim.
Take special care with numbers. Record what was counted, the denominator and the population described. In the fictional example below, 12 of 20 practice participants is 60% of that sample. It does not establish that 60% of all customers prefer the design. Keep the data source and context near the number so a reader can understand its limits.
Choose accurate wording or leave the claim open
Once you have checked the evidence, choose a defensible action: keep the supported statement, narrow its wording, explain an important condition or hold it for verification. If a claim adds little and cannot be supported, remove it rather than dress it in vague language. Missing evidence does not automatically prove a statement false; it means you have not established it.
Save a short record with the claim, original URL or document, relevant passage, access date, limits and decision. This makes a later update much easier. A simple evidence note is more useful than a bibliography full of links whose relationship to the wording is unclear.
- Supported: use wording that matches the evidence.
- Partly supported: narrow the scope or add the necessary condition.
- Unresolved: verify it before presenting it as a fact.
A small sample cannot speak for everyone
Fictional teaching data: 20 practice participants were asked to choose between two designs. Twelve chose A and eight chose B. The counts illustrate evidence scope; they are not a real survey, benchmark or business result.
| Draft claim | What the practice note supports | Editorial decision |
|---|---|---|
| All customers prefer A. | 12 of the 20 practice participants chose A. | Replace the universal claim with the sample-specific observation. |
| 60% of this practice sample chose A. | 12 ÷ 20 × 100 = 60%. | Keep with the fictional-data label and sample size. |
| Design A will increase sales. | The note records a preference choice, not purchases. | Do not present a sales effect as established. |
Put it into practice
Make a tiny claim ledger
A fictional draft uses the practice preference data above and also says that a product feature is available on every plan. The only feature document you have describes one named plan.
- Separate the preference, percentage, sales and plan-availability claims.
- Identify the evidence available for each statement.
- Rewrite any overbroad wording so it matches the actual evidence.
- Mark unsupported statements for removal or further verification.
- Record the source, its relevant scope and the reason for each decision.
Keep your working notes in your own document. This page saves only your completion check, not your answers or customer data.
Check your understanding
Make the call.
Choose an answer for each question. Then compare the reasoning.
You have a useful next step.
- Check specific statements instead of approving a whole paragraph at once.
- Read the original evidence and preserve its conditions and scope.
- Keep a clear record of supported, narrowed and unresolved claims.
Completion is your own learning record. Quiz scores and practice checks are saved separately, on this browser.
Sources & context
The sources below explain the factual foundations and editorial methods used in this lesson. Worked examples, planning models and exercises are illustrative teaching material.
- Carnegie Mellon University Libraries: Evaluate sources
Library guidance on source fit, lateral reading, checking the original claim and verifying AI-suggested citations.
- House of Commons Library: How to evaluate sources
Institutional explanation of relevance, methods, transparency and uncertainty. No UK legal discussion is carried into this lesson.
- UW–Madison: Vetting online content and fact checking claims
Academic professional-research guidance on tracing claims to original context. This lesson uses its own worksheet and fictional sample.