Turn question topics into a useful research plan.

Choose labels that describe a buyer’s decision.

Use a small set of labels you can explain to another person. Category discovery, agency workflows, and enterprise requirements describe different research needs. A label such as important tells you little about what the questions have in common.

For example, “Which tools track brand mentions in AI answers?” is a discovery question. “Which AI visibility tools support a small agency managing several clients?” introduces a specific audience and requirement. You can keep them in separate topics when you want to review those intents independently.

A topic is organizational context. It does not establish search demand or reveal how frequently real customers ask the questions. Use your own customer research to decide which intents deserve attention.

Check which questions the topic currently contains.

Echuno’s saved-answer Topic coverage view groups current, unarchived questions by their current topic labels. A question without a label appears under No topic assigned. Paused questions remain in this current set, even though they are excluded from future automatic collection.

Changing a topic label can regroup the question and its coverage in this view. Archiving a question removes it from the current topic set; its saved answers remain in the answer library. A changed topic total can therefore reflect library organization rather than new provider behavior.

Before comparing reports, record meaningful changes to the topic membership. If you edit the question itself, inspect the wording recorded with earlier answers. A stable question identity does not guarantee that every saved response came from identical wording.

Separate unanswered, unassessed, and unmentioned.

An unanswered question has no saved answer in the selected reporting scope. That is a collection gap. It does not establish that the provider omitted your brand, refused the question, or would never recommend the product.

A saved answer can still need brand assessment. Keep that state separate from an assessed answer that does not mention your brand. A failed or pending assessment is missing analysis, not a confirmed absence.

In Topic coverage, use Unanswered questions to find topics with collection gaps and Needs assessment to find topics where saved-answer counts exceed assessed-answer counts. Unmentioned questions identifies questions with assessments and no recorded mentions in that scope. Read individual answers to understand the result.

• No saved answer: review collection coverage and question state.
• Saved answer without assessment: review analysis status.
• Assessed answer without a mention: inspect the response and context.

Read answer counts alongside the mention rate.

A topic mention rate summarizes assessed answers, rather than giving every question equal weight. Questions sampled more often can contribute more observations. A topic with several questions is not automatically a balanced comparison.

Suppose one question has eight assessed answers with six mentions, while another has two assessed answers with no mentions. The combined rate is six of ten, or 60%. That does not mean 60% of the topic’s questions mention the brand: only one of these two questions has a recorded mention.

Check the questions, assessment counts, reporting period, and recorded platform coverage before comparing topics. A small sample or an uneven collection plan can make a headline percentage hard to interpret. A minimum assessed-answer filter helps narrow a review; it does not by itself make the sample representative.

Keep saved-answer coverage and API comparisons distinct.

The saved-answer topic view helps you find coverage gaps and review your brand’s recorded mentions. The separate API topic competitor report answers another question: how your brand and a chosen competitor appear within shared API answer samples.

In the API comparison, competitor-only answers are shared assessed samples mentioning the competitor without your brand. They are investigation leads, rather than proof that buyers prefer the competitor. Open the supporting evidence and compare the wording and recommendation context.

Do not silently combine browser captures and API responses into a single claim about what all users see. Likewise, a change across periods needs comparable questions, sample counts, and collection conditions. Topic labels alone do not make two samples equivalent.

Choose one concrete next investigation.

Start with the gap you can identify. If a relevant topic has unanswered enabled questions, inspect the collection plan and current queue before submitting more requests. If its questions are paused, decide whether collection should resume. A topic label does not schedule collection on its own.

If evidence exists, read a few answers from the questions behind the result. Look for a recurring requirement, a competitor recommendation, an ambiguous brand reference, or a cited page worth reviewing. Keep the saved response attached to your interpretation.

Record a specific follow-up and revisit the same question set after the change. A clearer product page or corrected third-party description may be worth doing on its own merits; it does not guarantee a future AI mention. Use the next saved samples to investigate what changed.