Not every source answers the same question
A specification page, a price, a direct test, a support policy and user feedback can each be useful. They should not be mixed into one undifferentiated “research” label. The evidence type determines what it can support and where caution is needed.
Use current official documentation for stated specifications and policies, direct testing for stated observations, credible third-party sources for context, and user feedback for patterns worth investigating. Date material evidence, record its limits and label unknowns. A claim without a source or test should not become a confident comparison point.
Match the evidence to the claim
| Evidence type | Useful for | Important limit |
|---|---|---|
| Official documentation | Declared specifications, compatibility, policies, support and published prices. | May be incomplete, region-specific or updated after publication. |
| Direct test or trial | Observed behaviour in a stated version, environment and task. | Cannot prove every user’s experience or long-term reliability. |
| Independent reporting | Context, corroboration and issues not covered by vendor materials. | Method, incentives, date and version still matter. |
| User feedback | Recurring experience, friction or questions to investigate. | May be fake, incentivised, atypical or based on an older version. |
| Unknown | Shows a decision-relevant gap. | Must remain visible; it is not a positive score. |
Review evidence for freshness and scope
- Check that the source refers to the exact product, version, plan, region or use case.
- Record the date checked and any limit that could affect a reader’s decision.
- Cross-check material claims rather than repeating one source as consensus.
- Keep a distinction between stated capability and observed experience.
- Correct or update a comparison when new evidence changes a material conclusion.
Reviews deserve context
The FTC advises readers to consider review source and recency rather than assuming a rating is reliable. See how to evaluate online reviews. A comparison should use user feedback to investigate a pattern, not as a substitute for a stated method.
Evidence FAQs
Can an official product page be trusted?
It is a good source for what a company currently states, but important details can be conditional, regional or incomplete. Pair it with the relevant policy, current terms and other evidence where the claim matters.
Are customer reviews evidence?
They can help identify a pattern or question, especially when recent and varied. They are not automatically verified, representative or comparable, and may need further investigation.
Why show an unknown instead of estimating?
An unknown can change a buying decision. Showing it honestly lets readers decide whether to investigate, wait or choose an option with clearer evidence.
Read comparisons with the right context
Use these guides to understand what a comparison can and cannot tell you. A useful result makes the inclusion, evidence, uncertainty, regional limits and commercial context visible before a reader acts.
