What brand sentiment tells you

KI Score shows how AI systems talk about your brand as its sentiment: in the Tonalität (Brand sentiment) column of the prompts table under Prompts, in the Sentiment card under Scoreboard → Overview, and in the Sentiment column under Responses. This article explains what the classification measures, how it is made, and why the bar is sometimes missing.

Note: The separate recommendation rate, formerly a column in the prompts table and a key figure in the Scoreboard, was removed on October 26, 2026. How often the AI recommends or praises your brand now shows up as the positive share of the sentiment.

What brand sentiment measures

A mention on its own doesn't tell you much. For potential customers, it makes a real difference whether an AI simply includes your brand in a list, explicitly highlights it as a good choice, or advises against it. Brand sentiment captures exactly that difference.

The tooltip on the Tonalität column describes it like this, in German: “Wie die KI Aussagen über deine Marke einordnet: positiv (Lob oder Empfehlung), neutral (sachliche Nennung) oder negativ (Kritik oder Abraten).” In English: how the AI classifies statements about your brand, positive (praise or a recommendation), neutral (a factual mention), or negative (criticism or advice against it).

So the base isn’t all answers, only the answers in which the analysis classified a mention of your brand. A positive share of 50% means that in half of these answers, your brand is recommended or praised. How often your brand is mentioned at all is shown by the Erwähnung % (Mention %) column in the same table.

How the classification is made

An AI analyzes the stored answers from your monitoring runs and identifies the brands they name. Your brand gets exactly one rating per answer:

• Positive: the brand is recommended, praised, or presented as a good choice.
• Neutral: the brand is only named or described factually.
• Negative: the answer advises against the brand, criticizes it, or presents it as a worse choice.

When in doubt, the analysis rates a mention as neutral. If an answer names your brand several times in different tones, the most negative rating counts. You can see the rating for each individual answer in the Sentiment column under Responses, which you show via Spalten (Columns).

A few more points about the calculation:

• The value covers the selected time period, such as Letzte 30 Tage (Last 30 days) or Gesamter Zeitraum (All time).
• The Gesamt (Total) row and the topic rows add up the classified answers of all prompts shown, so a prompt with more classified answers counts more.
• If you sort by Tonalität, the table orders the prompts by the positive share minus the negative share.
• The metrics are calculated with your current brand name and alias details. If you change them, the values for answers that are already stored change too.
• Brand sentiment is part of monitoring. It does not feed into the KI Score.

Why the bar can be missing

A missing bar is not a value of 0%. KI Score only shows the bar from five classified answers. Hovering over the cell shows, as n, how many there are so far. Typical reasons for too few classifications:

• In the selected period, there are fewer than five answers with a classified mention of your brand.
• None of the answers mention your brand.
• The prompt has no completed answers in the selected period yet, for example because no run has taken place.
• The analysis didn’t recognize your brand because the answers use a name you haven’t added as an alias.

Tip: Choosing a longer period brings more answers into the calculation. Otherwise, individual prompts with few mentions can swing sharply.

Following changes in sentiment

KI Score doesn’t offer a chart of sentiment over time or a separate score for it. The comparison period you pick with vs. Vergleich (vs. baseline) also shows changes only for the rates, the gaps, and the average position, not for sentiment. Here's how you can still see changes:

• In the prompts table, pick different periods one after the other, for example two months in the calendar, and compare the bars in the Tonalität column.
• In the Scoreboard, the Sentiment card under Overview shows the split into positive, neutral, and negative mentions for the selected period.
• Under Responses, use Filter hinzufügen → Sentiment (Add filter → Sentiment) to narrow the list to positive, neutral, or negative answers and read what the AI actually says.
• With a report, you get the split by email every week, month, or quarter. In it, the Sentiment block shows the mentions in the report's period. To learn how to set up a report and schedule its delivery, see Creating and emailing reports.