Turn a week of answers into a useful review.

Write down what this review covers.

Choose the project, reporting period, and providers before reading the numbers. In Echuno, Reports summarizes saved answers for the selected period and platform. Keep that scope visible when you copy or download the report.

If you compare this week with an earlier review, check whether the question set, tracked brands, wording, or collection method changed. A broader question set can change the result even when no individual answer became more favorable.

• Record the reporting period and which providers are included.
• Note additions, edits, or paused questions that affect coverage.
• Keep saved browser answers and API measurements clearly identified.

Check collection and assessment coverage first.

Use Tracking to see which questions have saved answers from each provider. A queued request, a failed attempt, and a saved answer are different states. An unanswered question does not show that a brand was missing from an AI response.

Then check brand assessments. Mention rates use assessed answers as their denominator. If some saved answers are still awaiting analysis, explain that gap beside the rate rather than treating the available assessments as a complete picture.

Read queue wait separately from answer collection time. Time spent waiting for the browser, other jobs, or a maintenance pause is not the provider’s answer-generation duration.

Choose a few observations worth reading.

Review a small number of concrete examples: a useful recommendation, a relevant question where your brand was absent, and a comparison that changed. Open the full saved responses and read the quotes behind their assessments.

Keep recommendation rank separate from mentions. An answer can mention a product without recommending it, and an unordered recommendation can have no explicit rank. An average rank describes only the answers with recorded positions.

For competitors, check assessment coverage before comparing mention rates. Two brands may have different denominators. A higher rate based on very few assessed answers deserves a narrower conclusion.

Trace a claim back to its saved sources.

Open Sources to find the websites and pages recorded with your answers, then follow View answers to read the surrounding context. A source link is a starting point for investigation; it does not prove why a provider selected a brand.

Look for inaccurate product details, unclear positioning, missing use cases, or outdated third-party descriptions. Verify the source yourself before assigning a content task. A frequently cited website is not automatically an endorsement opportunity.

• Observation: what the saved answer actually says.
• Evidence: the exact quote and any relevant saved source.
• Follow-up: the page or claim someone should verify.

Keep the team update short and traceable.

Use Copy report or Download report in Reports for a readable summary. Use CSV when someone needs to inspect the underlying counts. These outputs summarize saved data; they do not collect new answers or establish that every buyer sees the same response.

Add a short editorial note using the structure below. Link to the relevant saved answer when teammates have workspace access. If they do not, share an appropriate excerpt with the question, provider, collection date, and context needed to read it accurately.

• Scope: the project, dates, providers, and meaningful changes to the question set.
• Coverage: what was saved and assessed, and what remains missing or failed.
• Observations: two or three findings with supporting evidence.
• Actions: an owner, a concrete follow-up, and when to review it again.

Use the next review to revisit the evidence.

Keep the core question wording stable while following up on a finding. If you need to explore a different buyer need, add a separate question so you can interpret the two sets of results separately.

Record when a product or content change was made, then inspect later captures for repeated patterns. A change after publication does not establish that the publication caused it. Providers, models, sources, and answer wording can also change.