How you appear in AI answers

Review the questions measured, their visibility and the tone of mentions before interpreting the summary.

AI-answer visibility slides overview

This block includes scan-method and language explanations, query-stage visibility where classified prompts exist, a prompt table spanning as many slides as needed, and sentiment where the brand was mentioned. Explanatory examples carry Sample for illustration only, not your data; they are separate from your measured rows.

Report language, stored context language and prompt language are distinct. Read the actual question text to check the market and language being measured; do not assume every source or answer must use the PDF’s language.

Compare query stages and branded questions

By funnel stage groups questions into Awareness, Comparison & choice, Decision and, where needed, Not determined. Each group shows its prompt count, corpus share and average prompt visibility. If no prompt has a stage, this slide is omitted.

Branded queries names a tracked brand in the question, including a competitor. Unbranded queries does not name a tracked brand. This split differs from the stage classification.

Check how many and which prompts support each group. These are question categories, not tracked visitors moving through a sales funnel. A difference between groups does not by itself identify which technical or content change will help.

Read the per-prompt table

Read each question before its metrics. The table shows visibility, Tone, Competitors, Position, Sources and Category.

Visibility summarizes answers mentioning your brand. Position describes its average place where measured; a lower number means an earlier place in the answer, not a Google ranking. Sources counts distinct cited URLs for that prompt, rather than source domains. A dash indicates an unavailable value; do not read it as zero.

For the underlying answers, samples and scan history, open Prompts & answers. Edit the questions or their grouping in Prompts and categories. Saved PDF rows do not change after those edits.

Read sentiment without counting absence as neutral

Sentiment distribution groups prompts with brand mentions into five tone bands, from very positive to very negative. Its bars count prompts, rather than individual answers. Prompts with no brand mentions are excluded; with no mentioned prompts, the sentiment slide is omitted.

By model shows the breakdown by AI model. Strong prompts and Weak prompts list selected prompts with higher and lower tone; they are shortened lists. A model with no mentions has no measured mention tone.

Use Sentiment to examine the underlying mentions. Differences between models do not establish their cause or prove that the brand’s reputation changed outside the measured questions.