VERIFY ONLINE CLAIMS | Turn a Statement Into Questions You Can Actually Check

Learn how to break down viral posts, statistics, quotations,

news reports, and AI-generated answers into specific questions that can be investigated using evidence.

Why Good Research Begins Before You Search

A message appears in a WhatsApp group. A video makes a dramatic claim. Someone publishes a statistic without explaining where it came from. An AI assistant gives you a confident answer.

Your first instinct may be to search the exact wording, look for a familiar website, or ask another person whether the information is true.

But there is an important step before all of these: identify exactly what you are trying to establish.

A claim can contain several statements bundled together. One part may be correct while another is outdated, exaggerated, misleading, or unsupported. If you investigate the whole statement as if it were one simple fact, you can overlook those differences.

Digital research begins by converting the information you encounter into questions that evidence can answer.

This skill is useful when you are studying, writing a blog post, preparing a business report, checking a public announcement, comparing products, investigating a technical problem, or deciding whether to share information with other people.

The objective is not to become suspicious of everything. It is to become precise about what you believe, why you believe it, and what remains uncertain.

1. Understand What a Claim Is

A claim is a statement that asserts something about the world. Depending on its wording, it may be possible to support it, contradict it, qualify it, or determine that there is not enough evidence to reach a conclusion.

Consider these examples:

• “The company released a new application yesterday.”
• “This application is the safest option available.”
• “A survey found that 70% of customers prefer the service.”
• “The government has introduced a new requirement.”
• “This video shows an event that happened this morning.”
• “Using this method will double your income.”

These statements do not require the same investigation.

The first is a question about a release and its date. The second makes a comparative safety claim that requires defined criteria and evidence. The third requires a survey and enough information to understand how the result was obtained. The fourth needs an identifiable official announcement or legal document. The fifth requires evidence about the video's origin, date, and context. The last makes a prediction or performance claim that needs much more than a testimonial.

The wording tells you what kind of evidence to look for.

Separate facts, interpretations, opinions, and predictions

Before researching a statement, identify what sort of statement it is.

A factual claim asserts something that can potentially be checked against records, observations, measurements, or other evidence.

Example: “The application was released on 12 September.”

An interpretation explains what someone thinks a fact or set of facts means.

Example: “The release shows that the company is preparing to enter a new market.”

An opinion expresses a judgment or preference.

Example: “This is the best application for beginners.”

An opinion can be informed by evidence, but the word best has no precise meaning until you identify the criteria being used.

A prediction describes what someone expects to happen.

Example: “The application will become the market leader next year.”

A prediction cannot normally be verified as a completed outcome before the relevant future period has passed. You can, however, examine its assumptions, supporting evidence, and plausibility.

Many online messages combine all four. Your task is to separate the parts so that each can be investigated appropriately.

2. Break Compound Claims Into Smaller Questions

Imagine receiving this message:

“Starting next month, every small business in Nigeria must register on a new government platform, or it will be prohibited from operating. The registration is free and takes only five minutes.”

For this exercise, the message is hypothetical. It is not an assertion that such a policy exists.

It contains several distinct claims:

1. A new government platform exists.
2. A registration requirement will begin next month.
3. The requirement applies to every small business in Nigeria.
4. Businesses that do not register will be prohibited from operating.
5. Registration is free.
6. Registration takes five minutes.

These statements require different evidence. A platform might exist without being compulsory. A requirement might apply only to particular business categories. A registration process might be free while a related service carries a charge. A stated deadline might have changed.

If you search only for the complete message, you may find people repeating it without establishing any of its individual parts.

Instead, turn each part into a question.

Claim	Research question	Evidence to seek
A platform exists	What is the platform's official name and who operates it?	Official portal or announcement
Registration begins next month	What is the effective date?	Dated notice, regulation, or official guidance
All small businesses are covered	Which businesses and legal categories are included?	Applicable rules and definitions
Non-registration prevents operation	What consequence is explicitly specified?	Relevant legal provision or official guidance
Registration is free	Are there official charges or third-party fees?	Current fee schedule and official instructions
Registration takes five minutes	What is the basis for this time estimate?	Documented process or measured experience

Notice the difference: the original message invites you to accept or reject a dramatic story. The questions make it possible to investigate individual details.

Rule: When a claim contains multiple factual assertions, split it into smaller claims before deciding whether the whole statement is accurate.

3. Make Each Question Specific Enough to Investigate

A question such as “Is this true?” is often too broad to guide useful research.

Improve it by identifying the subject, action, time period, location, and conditions that matter.

For example:

Broad question: Is the policy real?

More useful questions:

• Which institution issued the policy?
• What is the official title or reference number?
• When was it announced?
• When does it take effect?
• Which people or organizations are covered?
• Is the measure mandatory, recommended, or proposed?
• Has it been amended, postponed, or withdrawn?
• What document establishes the consequences of non-compliance?

You will not need every question for every claim. Select the details that could change the answer.

Use the who, what, when, where, why, and how framework

These prompts help you expose missing details:

• Who? Who made the claim, issued the statement, conducted the research, or is affected?
• What? What exactly happened or is being asserted?
• When? When did the event occur, and when was the information published?
• Where? What location, jurisdiction, organization, or population does the claim concern?
• Why? What explanation or cause is being proposed, and what evidence supports it?
• How? How was the result measured, the event recorded, or the conclusion reached?

For a technical claim, add questions about the device, software version, configuration, test conditions, and measurement method.

For a health or financial claim, identify the population, timeframe, outcome, risks, and conditions under which the claim is supposed to apply. Do not treat a general online claim as a substitute for qualified professional advice in high-stakes situations.

4. Distinguish the Claim From the Evidence

A repeated statement is not automatically evidence for itself.

Suppose ten social media accounts repeat the same statistic. At first glance, the agreement may appear convincing. But if all ten accounts copied the same unverified post, you may have only one original source—and possibly no reliable evidence at all.

Separate three things:

The claim: What someone says is true.

The evidence: The records, measurements, documents, observations, or other material that support or challenge the claim.

The inference: The conclusion you draw from the evidence.

Consider a simple example.

A post says that a particular website is the most visited website in a country. It includes a screenshot of a traffic ranking.

• The claim is that the website ranks first.
• The screenshot is evidence that someone displayed a ranking at a particular time.
• The inference that the website is currently the most visited depends on the ranking's source, date, methodology, geographic scope, and definition of visits.

The screenshot alone does not establish all those details.

A useful research question is therefore not merely “Can I find this statement elsewhere?” It is “What evidence would justify this statement, and can I inspect that evidence?”

5. Decide What Would Count as a Satisfactory Answer

Before opening a search engine, think about what would actually settle—or meaningfully narrow—the question.

Different claims call for different evidence.

Type of claim	Useful evidence to seek
A law or government requirement exists	Applicable legislation, official regulations, notices, or authoritative government guidance
A person made a quotation	Original recording, transcript, complete speech, or a reliable publication close to the source
A statistic is accurate	Original dataset or report, definitions, sample, timeframe, and methodology
A software feature exists	Current official documentation and, where practical, a test of the feature
A scientific finding is supported	Relevant research, methods, limitations, subsequent evidence, and suitable expert analysis
A photograph depicts a particular event	Original source, earlier appearances, date and location evidence, and independent corroboration
A business achieved a claimed result	Verifiable records or transparent methods, with enough context to judge the result

This is not a rigid ranking in which one source type always wins. An official document is especially relevant to what an institution formally announced, but it may not independently establish whether a factual assertion in that document is correct. A scientific paper may be directly relevant to its research question while still having methodological limitations.

The right question is: Is this evidence suitable for the particular claim I am evaluating?

6. Turn the Questions Into a Search Plan

Once you have isolated the claims and defined the evidence you need, build a small search plan.

Start with the most identifiable details:

• Exact names and quoted wording.
• Dates, document titles, reference numbers, and locations.
• The name of the institution, author, company, or research group.
• The type of evidence you expect to find.
• Alternative wording that the original source might use.

For the hypothetical registration message, a search plan could look like this:

Search A — Identify the institution and platform.

Search for the claimed platform name alongside the institution said to operate it.

Search B — Find the official announcement.

Search for the key phrase alongside terms such as official notice, registration requirement, effective date, or guidance. These are search terms, not proof that a notice exists.

Search C — Establish the scope.

Look for the document's definitions, covered categories, exclusions, and jurisdiction.

Search D — Check the claimed consequences and fees.

Search the relevant official documents for the stated penalty, fee, or deadline. Confirm that the document applies to the situation being investigated.

Search E — Look for corrections or later updates.

Check whether the original announcement has been clarified, amended, postponed, or superseded.

Do not assume the first result is the best result. Search results are leads, not verdicts. If the first search does not identify the original evidence, revise the query rather than repeatedly searching the same wording.

You can use different search engines, library catalogues, scholarly indexes, official databases, or specialist research services depending on the question. The research method should determine the tool—not the other way around.

7. Record What You Know and What You Still Need to Find

Research becomes unreliable when you remember a conclusion but forget how you reached it.

Create a simple research log in a notebook, text document, spreadsheet, or note-taking application. No special paid software is required.

Use this template:

Field	Your entry
Claim ID	A short identifier, such as C01
Exact claim	The wording you are investigating
Sub-question	One specific question to answer
Evidence required	What would help establish the answer
Search performed	The query or method used
Source	Title, author or institution, and link
Dates	Publication date and date accessed
Relevant finding	What the source actually establishes
Limitations	Missing context, uncertainty, or conflicts
Status	Unchecked, supported, challenged, or unresolved
Next action	The next search or verification step

Keep source findings separate from your own interpretations. If a document says a policy is proposed, do not record that as proof that the policy is already in force.

If a source cannot be found, record that too. Failure to locate a document is not always proof that it does not exist. The document may use different wording, be hosted elsewhere, or be unavailable to the public.

The purpose of the log is to make your investigation reproducible. Another person should be able to follow your steps, inspect your sources, and understand why you reached your conclusion.

8. Know When to Stop and When to Investigate Further

Not every claim deserves the same amount of research.

A casual question about a minor software setting may need only a current official help page and a quick test. A claim about a legal obligation, a medical treatment, a major financial decision, or a public safety emergency requires a higher standard of evidence and, where appropriate, a qualified authority.

Use three practical questions:

1. What happens if I am wrong? The greater the potential harm, the more careful the investigation should be.
2. What evidence do I have? A confident tone, a popular post, a screenshot, or repeated copies may not be sufficient.
3. What remains uncertain? If a missing detail could change the conclusion, do not hide it.

A reasonable stopping point is reached when you have answered the important sub-questions with evidence appropriate to the stakes, checked meaningful counterevidence, and recorded any remaining limitations.

Sometimes the honest result is “supported by the available evidence.” Sometimes it is “contradicted by the available evidence.” Sometimes the correct result is “unresolved.”

An unresolved claim is not a failed investigation. It is a conclusion that accurately reflects the limits of what you could establish.

9. Practice: Investigate a Claim Without Assuming It Is True

Use this invented statement for practice:

“A new free mobile application can recover every deleted file from any phone in five minutes.”

Do not assume the statement is true or false. Decompose it.

Step 1 — Identify the sub-claims.

• The application exists.
• It is free.
• It can recover deleted files.
• It works on every phone.
• It can recover every type of file.
• It works within five minutes.

Step 2 — Define the evidence needed.

You would need to identify the actual application and developer, review the supported devices and operating systems, examine the pricing and limitations, and find credible testing or technical documentation about recovery success.

Step 3 — Identify the hidden conditions.

Recovery may depend on whether data has been overwritten, the phone's storage technology, backups, operating-system restrictions, and the type of file. A demonstration on one device would not establish universal compatibility.

Step 4 — Build a search plan.

Start with the exact application name, developer, official documentation, supported-device list, privacy policy, and independent technical evaluations. If the application name is missing, that itself is a reason to seek clarification before trusting the claim.

Step 5 — Record the result.

Do not mark the claim “true” simply because an application exists. Each important part needs its own evidence.

Your deliverable is a short research record containing the original statement, at least four testable sub-claims, the evidence required for each, your search plan, and the questions that remain unresolved.

10. Your Completion Check

Before considering this skill learned, make sure you can:

• Separate factual claims from interpretations, opinions, and predictions.
• Break a compound statement into smaller, testable claims.
• Identify the specific details that could change an answer.
• Describe what evidence would be suitable for each question.
• Build a search plan instead of relying on one broad query.
• Record sources, findings, dates, limitations, and unanswered questions.
• Explain why an unresolved claim should not be presented as a verified fact.

If you can do these things, you have established the first part of a dependable research method.

Conclusion: Precision Comes Before Proof

Online research is not simply the ability to find information quickly. It is the ability to identify what needs to be established, find evidence suited to that question, and communicate the result without claiming more than the evidence supports.

The next time you encounter a dramatic statistic, an unfamiliar announcement, a technical promise, or a confident AI-generated answer, resist the urge to accept or reject the whole statement immediately.

Break it down. Define the questions. Decide what evidence you need. Then investigate.

That small change turns casual searching into a disciplined research process.

Further Reading

• Stanford Graduate School of Education — Civic Online Reasoning: https://ed.stanford.edu/node/21844
• UNESCO — Media and Information Literacy: https://www.unesco.org/en/media-information-literacy
• University of Michigan–Dearborn — SIFT: Evaluating Web Content: https://guides.umd.umich.edu/SIFT

These resources provide additional context on evaluating online information and investigating sources. They complement the method taught here; they do not replace the need to examine the evidence for each individual claim.

W3 NOTES BLOG | The Internet Money Feed 

Image: The Balance Money