AI share of voice is the fraction of buyer recommendation prompts where your brand gets named, and cited, by AI answers, verified against the exact sentence each engine used. If you cannot check that sentence and source, you are working from assumptions. Yuzu is built for that measurement style: it records whether your brand is named, what the ranked alternative field looks like, and the sentence each answer came from.
What people mean by AI share of voice
Most people use “share of voice” as shorthand for visibility. With AI answers, the signal is more direct: when a buyer asks for a recommendation, do the answer engines name your brand, or do they name someone else?
According to the Yuzu homepage, Yuzu frames the measurement around that moment:
- whether buyer questions in your category get your brand named in AI answers
- the ranked field, which brands the engines recommend instead
- citations, when an answer links your site, not just names it
- search visibility alongside, from Google Search Console
That framing makes the term easier to audit because you can look at what the answer actually says. If you are reviewing a recommendation prompt from your own category, Yuzu also keeps the reading check linkages, so every number links to the answer text it was read from and the reading can be checked.
Why AI share of voice is not a vanity metric
If you only track impressions or follower counts, you still do not know whether AI surfaces your brand during buyer decisions.
According to the Yuzu materials, the approach is built around evidence:
- it measures whether ChatGPT, Perplexity, and Google name your brand when your buyers ask for a recommendation
- it records the ranked field when they do not name you
- it captures the exact sentence each reading came from
In practice, that gives you something you can point to. If a buyer asks for tools in your category, the result is: your brand was named here, and these were the sentences and the alternatives.
The measurement inputs: prompts based on real buyer questions
The hardest part of AI visibility measurement is prompt quality. If your prompts are not close to what buyers actually ask, you end up measuring the wrong thing.
According to the Yuzu homepage, it derives the questions from your own homepage. The flow is:
- you add your domain
- Yuzu derives the questions your buyers ask from your own homepage
- it puts those questions to the answer engines
- it records who was named, you or the ranked field of competitors named instead
Because the questions come from your own site, the measurement tries to stay aligned with the way your offering is presented. If your homepage says one thing and the answers reflect another, that is the kind of mismatch you want to catch. And because every reading links back to the sentence, you can verify what happened instead of accepting a dashboard number.
Named vs cited: the split to track
A lot of teams treat “mention” as success. In day to day reporting, that leaves out a useful distinction.
Yuzu explicitly tracks two related but different outcomes:
- named: the engine includes your brand in the answer
- cited: when the answer links your site, not just names it
That difference matters when you are checking actual traffic paths. An answer can name you while sending the reader nowhere near your site. With citations tracked as a separate outcome, you can see whether your content signals are strong enough to earn the link, not only the mention.
AI share of voice: what to measure side by side
You get a clearer read if you measure AI naming and citation together, and keep search visibility alongside them. According to the Yuzu homepage, Yuzu also includes search visibility alongside, from Google Search Console.
Here’s the clean comparison:
| Metric type | What it answers | Yuzu records it as | |---|---|---| | Named in AI answers | Did the engine include your brand in its recommendation? | Whether buyer questions get your brand named | | Ranked field | If it didn’t name you, who did it recommend instead? | The ranked field: which brands the engines recommend instead | | Citations | Did the engine link to your site, not just name it? | Citations, when an answer links your site | | Search visibility alongside | Are you seeing corresponding search signals in Google? | Search visibility alongside, from Google Search Console |
If you are comparing weekly results, this keeps the categories separate. You can tell the difference between being mentioned, being cited, and being displaced by another brand in the ranked field.
How to check AI share of voice without guessing
The biggest failure mode in AI visibility work is relying on made up measurements. Yuzu’s own blog materials emphasize sentence level verification.
For example, the product documentation points to a specific blog post: “How to find out what ChatGPT and Perplexity say about your brand,” and it positions that as a manual, free check where you can tell a real measurement from a made up number. That logic fits AI share of voice work more broadly:
- find the exact sentence AI used
- compare it to what you would expect from your published signals
- verify whether your site was linked or only named
- review the alternatives, so you know what “share” you are actually losing
If you want a practical path, you can start with the existing guides:
- How to find out what ChatGPT and Perplexity say about your brand
- How to Get Mentioned by ChatGPT: A Practical Checklist for Real Visibility
How AI share of voice connects to content decisions
Once you accept the measurement terms, content decisions get more specific. The goal becomes: produce the signals the answer engines can cite.
According to Yuzu’s LLMs and accuracy notes, Yuzu describes content discipline checks as part of its growth engine. The engine includes:
- ingest
- fetch, probe, audit, draft, publish
It also runs an “audit” step and “content discipline checks,” plus a “doctor” step that learns lessons as a rule. In practice, that means measurement informs what to publish next, and the reading can be checked by linking back to the sentence each reading came from.
If you are trying to reason about brand naming behavior across engines, the blog also provides brand recommendation comparisons and brand naming checks, including:
- ChatGPT vs Perplexity Brand Recommendations: Which Names You
- Why is my brand not mentioned in ChatGPT?
Why AI share of voice needs evidence in the record
When your team, clients or prospects ask “What did AI actually say?”, you need more than a dashboard claim. You need the sentence.
According to the Yuzu homepage, the model is built around that idea:
- you measure whether ChatGPT, Perplexity, and Google name your brand
- you can check the exact sentence each reading came from
- every number links to the answer text it was read from
That is what turns AI share of voice from a marketing phrase into something you can defend.