Follow your curiosity.

Separate the contribution from the confident claims.

Know which claims deserve a closer look.

Find patterns without flattening disagreement.

Compare outputs against the same job.

Bound the question before collecting more tabs.

See which claims are supported and which are not.

Trace the denominator, date, and original source.

Separate the sample result from the wider population.

Find the reason two accounts disagree.

Learn in a useful sequence, with checkpoints.

Make interpretation explicit and repeatable.

Avoid citations that merely mention the topic.

Prioritise the claims most likely to mislead.

Separate promotional language from findings.

Find the denominator, period and population.

Make the caption match what the chart shows.

Keep conclusions within the study design.

Find authoritative evidence for a bounded question.

Preserve qualifications that affect interpretation.

Use interview time to resolve genuine uncertainty.

Find whether the figures are actually comparable.

Spot facts likely to have changed.

Focus on evidence that changes an action.

Improve neutrality before collecting responses.

Make comparison criteria explicit before choosing.

Keep vivid examples in proportion.

Be precise about what is not known.

Avoid counting repeated reporting as independent support.