Metrics

A practical reference for the metrics Voxoria tracks: how AI engines talk about your brand, your competitors, and the sources shaping those answers.

Each metric can be filtered by date, engine, topic, country, prompt, and other reporting slices where the product supports it.

Quick reads

Need a board metric? Use visibility and share of voice. They explain whether the brand is present and how strongly it competes.

Need a content priority? Use prompt visibility, average position, and source retrieval to see which topics and pages need work.

Need a brand narrative read? Use sentiment, endorsed claims, and undermined claims to understand what AI is teaching buyers.

Brand visibility

Visibility

What it is: The percentage of tracked AI answers that mention your brand at least once.
Why it is useful: This is the simplest read on whether AI engines know to include you when buyers ask category questions.
Example: If Voxoria runs 200 answers this week and 80 mention Acme, Acme’s visibility is 40%.

What it is: Your brand’s mentions as a share of all tracked brand mentions across the answers.
Why it is useful: Visibility says whether you appear. Share of voice says how much of the conversation you own versus competitors.
Example: If AI answers mention tracked brands 1,200 times and 360 of those mentions are Acme, Acme has 30% share of voice.

Mentions

What it is: The total number of times your brand is named in the answers we track.
Why it is useful: Useful for spotting volume shifts. A campaign may not lift visibility immediately, but it can increase repeat mentions inside answers.
Example: Acme is mentioned 540 times across 1,200 answers this month.

Average position

What it is: The average order your brand appears in when AI lists or compares brands. Lower is better.
Why it is useful: Being named first usually signals stronger association with the category than being buried after several competitors.
Example: If Acme is first in some answers and third in others, its average position might be #1.8.

Sentiment and preference

Sentiment

What it is: A 0 to 100 score based on whether AI describes your brand positively, neutrally, or negatively.
Why it is useful: A brand can be visible for the wrong reasons. Sentiment shows whether mentions are helping or hurting buyer perception.
Example: If Acme has 12 endorsed, 5 neutral, and 3 undermined statements, the score is 72.5 out of 100.

Endorsed, neutral, undermined

What it is: Counts of the specific claims AI makes about a brand, grouped by whether they help, merely describe, or work against it.
Why it is useful: The split tells you what to fix. A low score from neutral mentions needs different action than a low score from repeated criticisms.
Example: An answer saying Acme is easy to adopt is endorsed; saying Acme exists is neutral; saying Acme lacks enterprise controls is undermined.

Net sentiment

What it is: A brand-only polarity score from -1 to +1, comparing endorsed statements against undermined statements.
Why it is useful: Good for tracking whether your brand narrative is improving, independent of whether competitors are moving too.
Example: If Acme has many more endorsed than undermined claims, net sentiment moves toward +1.
Same signal, two scales: Net sentiment and the 0–100 Sentiment score above read the same endorsed / neutral / undermined data. The dashboard’s Sentiment tile shows the 0–100 version; the Sentiment page leads with net sentiment (-1 to +1). Don’t be thrown by seeing both numbers — they’re two views of one thing.

What it is: Your brand’s contribution to the workspace’s net positive sentiment across all tracked brands.
Why it is useful: Useful for competitive reporting because it asks: of the positive AI preference in this market, how much belongs to us?
Example: If Acme has 9 more endorsed than undermined claims in a market with 60 total sentiment events, Acme owns +15% net endorsement share.

Sources and citations

Citations

What it is: The number of cited URL rows AI answers used in the selected period.
Why it is useful: Citations show which pages AI engines treat as evidence, not just which brands they mention.
Example: If AI cites 540 source URLs this week, those citations reveal which domains are influencing the answers.

Retrieved

What it is: The percentage of answers where a domain or URL appeared as a source, whether or not it was explicitly cited.
Why it is useful: A page can influence an answer before it gets cited. Retrieval is an early signal that AI engines are using your content.
Example: If acme.com appears as a source in 40 of 100 answers, its retrieved rate is 40%.

Citation rate

What it is: The average number of citations a domain or URL earns when it is used as a source.
Why it is useful: Helps separate pages that are occasionally discovered from pages that AI engines repeatedly lean on as supporting evidence.
Example: If Acme is cited 60 times across 40 cited answers, its citation rate is 1.5.

Source mix

What it is: A breakdown of cited sources into you, competitors, related sources, and unknown domains.
Why it is useful: Shows whether AI answers are being backed by your owned content, competitor content, or third-party sources.
Example: A healthy source mix might show 45% of citations from your domain, 30% from competitors, and 25% from related publishers.

Channel and prompt views

What it is: A channel’s share of total brand mentions, split by engines such as ChatGPT, Claude, Gemini, Perplexity, Copilot, and Google AI Overviews.
Why it is useful: Different engines can tell different market stories. Channel share shows where the conversation is happening.
Example: ChatGPT might produce 40% of tracked mentions while Claude produces 35% and Gemini produces 25%.

You by channel

What it is: The percentage of answers on a specific engine that mention your brand.
Why it is useful: Shows where to focus. You may perform well in ChatGPT but be almost absent in Google AI Overviews.
Example: If 32 of 80 ChatGPT answers mention Acme, Acme’s ChatGPT visibility is 40%.

Prompt visibility

What it is: Your visibility for a specific buyer question or topic group.
Why it is useful: Turns a broad score into an action list by showing which prompts you win, lose, or need to build content for.
Example: Acme may have 65% visibility for ‘best CRM for agencies’ but only 10% for ‘CRM with AI lead scoring’.

Top brands

What it is: The brands most often mentioned for a prompt, topic, or market slice.
Why it is useful: Useful for competitive planning because it shows who AI engines associate with each buyer need.
Example: For ‘best project management tool for engineering teams’, the top brands might be Linear, Asana, Jira, ClickUp, and Notion.

How to read movement

Look for direction, volume, and evidence. A stronger AI search story usually shows up in three ways: more answers mention you, those mentions move earlier and become more positive, and AI engines start using your own pages as evidence.

No single number explains the whole channel. Visibility tells you if you are present. Share of voice tells you if you are winning. Sentiment tells you whether the answer helps. Source metrics tell you what to change next.