Negative isn’t the same as wrong.
Every claim gets a framing. Whether it’s true is your team’s call: record a correction and it’s kept apart from claims that only read negative.
See what AI says about you. Then mark what it gets wrong.
Perception asks AI about your brand, splits every answer into claims and keeps how they read apart from whether they’re right.
Are YourBrand running shoes worth the price?
YourBrand makes cushioned daily trainers for road running. Runners often praise the Glide line for comfort over long distances. Its shoes are expensive for what you get. YourBrand doesn’t offer free returns, so buying the wrong size can be costly. Compared with Stridewell, it has fewer options for wide feet. If price matters most, Stridewell or Fleetfoot may be better value.
4 claims found
ClaimFramingYour review
Runners often praise the Glide line for comfort over long distances.PositiveNot reviewed
Its shoes are expensive for what you get.NegativeNot reviewed
YourBrand doesn’t offer free returns, so buying the wrong size can be costly.NegativeCorrection recorded
Compared with Stridewell, it has fewer options for wide feet.NegativeReviewed
Framing shows how a claim reads, not whether it’s true. Your team’s review records that.
Five kinds of question. Asked of each AI platform you track.
Questions that name you, worded the same every run. Buyer prompts are tracked separately in Analytics.
Sample answers
- ChatGPTA solid pick for long, steady miles. 78Positive
ClaudeWell reviewed for cushioning at high mileage. 72Positive- GeminiPopular, though fit varies by model. 54Neutral
Microsoft CopilotPricier than similar trainers. 31Negative- Google AI OverviewsLists running brands without assessing YourBrand. –Said nothing clear
- ChatGPTSimilar cushioning; Stridewell fits wider feet. 49Neutral
- ClaudeStridewell offers more wide-fit models. 24Negative
- GeminiYourBrand feels lighter on long runs. 81Positive
- Microsoft CopilotBoth are good; it depends on fit. 57Neutral
- Google AI OverviewsYourBrand is often rated higher for comfort. 74Positive
- ChatGPTStridewell, Fleetfoot and Northpeak. 51Neutral
- ClaudeFleetfoot for value, Stridewell for wide feet. 46Neutral
- GeminiCheaper shoes give similar comfort. 28Negative
- Microsoft CopilotRepeats the question without an answer. –Said nothing clear
- Google AI OverviewsStridewell and Fleetfoot are common picks. 54Neutral
- ChatGPTExpensive for what you get. 22Negative
- ClaudeWorth it if you run high mileage. 69Positive
- GeminiNo free returns makes sizing risky. 35Negative
- Microsoft CopilotSimilar shoes cost less. 31Negative
- Google AI OverviewsDepends on how often you run. 49Neutral
- ChatGPTNo, returns are not free. 24Negative
- ClaudeCheck the retailer’s returns policy. 57Neutral
- GeminiReturns cost a fee. 28Negative
- Microsoft CopilotYes, within 30 days. 78Positive
- Google AI OverviewsNo clear answer in the overview. –Said nothing clear
From answer to action. One claim at a time.
Trace any claim to its source.
Open the answer behind a claim and the pages it cited. A citation shows what the answer linked to, not where the sentence came from.
Trace a claim back to its answer
Review it, then fix the cause.
Flagged claims are grouped into subjects with their evidence label. Record your review, then plan the fix with Sira: update your own content, or check a cited page before outreach. A review doesn’t change the AI answer.
Review what AI gets wrong about your brand
Numbers you can read honestly. Context shown, not hidden.
How AI frames you, in one score.
0–100 across scored answers, where 50 is neutral. Each platform counts equally, and the sample is shown beside it.
Negative isn’t the same as wrong.
A claim can read negative and still be true. Only your team’s correction marks it wrong.
A snapshot until runs compare.
A trend needs several runs with the same questions and platforms.
Coverage is part of the result.
A platform with too few usable answers is shown, not quietly dropped.
Your review is a record.
Marking a claim reviewed or recording a correction changes neither the AI answer nor the score.
Questions about Perception. Answered plainly.
What is Videntic Perception?
Perception is the part of Videntic that checks what AI platforms say about your brand. On every run it asks each AI platform you track the same declared questions that name your brand, across five types: brand quality, comparison, objection, alternative and facts. It splits each answer into claims and shows how each one frames you. You can trace each claim to the answer and sources behind it, record your team’s review and plan a fix with Sira.
How is Perception different from AI visibility tracking?
Analytics tracks buyer prompts that don’t name you and measures whether you are mentioned or cited. Perception asks declared questions about your brand by name and shows what the answers claim about you and how they frame you.
Does Perception know whether a claim is true?
No. Perception measures how answers frame your brand; it doesn’t fact-check claims on its own. Your team marks a claim reviewed or records a correction, and that review is kept beside the measurement without changing the score.
If an answer cites a source, does that prove the claim?
No. A citation shows what the answer linked to, not where a specific sentence came from. Open the answer and the source and check them before you act.
What does the Perception score depend on?
The platforms that returned enough usable answers in that run, each weighted equally, and the number of answers scored. On the 0–100 scale, 50 is neutral. An AI model splits each answer into claims and labels how each one reads: positive, neutral or negative. Answers that said nothing clear are set aside. A single run is a snapshot; a trend needs four comparable runs with the same questions and platforms.
Will Perception improve how AI describes my brand?
Not by itself. It shows what answers say and where to look. Changes come from the content and sources you fix, and no one can guarantee how an AI platform will answer.
Related guides
Know what AI tells buyers about you. And where it’s wrong.
Run your declared questions, read the claims and record what needs fixing.
Start free trial
7-day free trial.