The AI Reality Score
The AI Reality Score is Zicy’s per-engine accuracy metric, a score from 0 to 100 measuring how accurately AI engines describe a brand across its tracked prompt set, where 100 means every tracked response was factually correct.
A mention is only worth as much as the claim attached to it.
Most AI visibility tools measure presence: was the brand named? The AI Reality Score measures something harder and more important: was what AI said actually true? A brand can appear in nine out of ten AI answers and still be described with an outdated pricing tier, a wrong founding year, or a product feature it discontinued. That inaccuracy reaches the buyer before any page on the brand’s website does.
The AI Reality Score exists because a mention is only valuable if the claim attached to it is correct. It is the metric that turns AI visibility from a reach measure into a truth measure, and the number that tells a brand whether its AI visibility work is making AI more accurate, not just more frequent.
Four signals in every answer.
Accurate descriptions
AI responses that correctly describe the brand’s category, offering, positioning, key facts, and current state. These lift the score.
Outdated claims
Statements that were once true but are no longer: old pricing, former team members, discontinued products, superseded messaging. These reduce the score.
Hallucinations
Statements AI asserts about the brand that have no grounding in any published source and were never true. These reduce the score more heavily than outdated claims.
Entity gaps
Facts the brand should be known for that AI consistently omits: a key product line never mentioned, a market never attributed, a credential never surfaced. These indicate incomplete entity definition and reduce the score.
Scored on what AI says, per engine, over time.
The AI Reality Score is derived from Zicy’s structured analysis of actual AI responses across the brand’s tracked prompt set on each engine. For each prompt, the AI response is assessed against the brand’s known facts: what is stated accurately, what is outdated, what is false, what is missing.
Important: AI-generated responses are probabilistic. The same prompt asked at different times, or phrased slightly differently, can produce different outputs. The AI Reality Score is therefore a directional indicator tracked over time, not a single definitive measurement. Zicy tracks it weekly so changes are meaningful trends, not noise.
A score of 100 means every assessed response across the tracked prompt set was accurate on that engine. Scores below 100 reflect the proportion and severity of inaccuracies found. Scores are per-engine: a brand may score 94 on ChatGPT and 71 on Gemini (illustrative figures), reflecting genuine differences in how each engine has ingested and represented the brand’s entity data.
Inaccuracy, in weight order.
- Hallucinations (highest weight).
- Outdated claims.
- Wrong attributions (features attributed to the wrong product, facts attributed to the wrong team member).
- Missing key facts consistently absent across responses (entity gaps).
- Competitor conflation (AI describing the brand with a competitor’s attributes).
Give AI clear, correct, corroborated facts.
Schema and llms.txt
Make the brand’s correct facts machine-readable and unambiguous. The Action Center in Zicy generates both.
Content accuracy
Publish and maintain clear, crawlable, authoritative content on the facts AI gets wrong. The Content Optimizer identifies which existing pages need updating.
Entity gap remediation
Create content for the topics AI consistently omits. The Visibility Gaps module prioritises these by estimated impact.
Source corroboration
Earn citations from independent sources that state the correct facts. This is the earned-media layer, what AI Citation Enablement is designed to build.
See the modules on the platform, or ask the AEO/GEO consultant what to fix first.
Accuracy reduces hesitation at the moment of decision.
An accurate AI description reduces buyer hesitation. When AI describes a brand correctly, right category, right differentiators, right proof, the buyer who acts on that answer arrives with correct expectations and higher purchase confidence. Zicy’s Site Traffic module connects AI Reality Score improvements to branded-search behaviour and AI-referred sessions over time, the Prove layer of the loop.
Questions about the AI Reality Score.
What is the AI Reality Score?
The AI Reality Score is Zicy’s per-engine accuracy metric, a score from 0 to 100 measuring how accurately AI engines describe a brand across its tracked prompt set. It is Zicy’s core differentiator: most tools measure whether a brand was named; Zicy also measures whether what was said was true.
How is the AI Reality Score different from a brand mention score?
Brand mention coverage tells you how often your brand appeared in AI answers. The AI Reality Score tells you whether those appearances described you correctly. A brand with 90% mention coverage and a Reality Score of 58 is being named frequently but described inaccurately, which may be worse than not being named at all.
Why does my score differ between engines?
Each AI engine has ingested different sources, applies different retrieval weights, and updates its knowledge at different rates. Gemini may have indexed a press release that ChatGPT has not. Perplexity may be citing a source with outdated information that the others have deprioritised. Scores differ between engines because the engines themselves differ in what they know about your brand.
Can I improve my AI Reality Score?
Yes. The most reliable improvements come from making correct facts machine-readable (schema, llms.txt), publishing and updating accurate content on the specific claims AI gets wrong, filling entity gaps with new content, and building corroborating citations from independent sources. Zicy’s Action Center, Content Generator, and Visibility Gaps module are built for exactly this.
How often is the AI Reality Score recalculated?
On Starter and Growth plans, Zicy tracks prompts weekly and updates scores accordingly. On Enterprise plans, daily tracking is available. Because AI responses are probabilistic, weekly tracking gives directional trends rather than single-point readings.
Is a score of 92 out of 100 good?
It depends on what is pulling it below 100. A score of 92 with one minor outdated claim is a very different situation from a score of 92 with a hallucination that misattributes a product feature. Zicy’s Brand Intelligence report breaks down exactly what is reducing the score so the fix can be targeted.
What counts as a hallucination in Zicy's scoring?
A hallucination is a specific factual claim AI makes about a brand that has no grounding in any published source and was never true, a wrong founding year, a non-existent product, a team member who never worked there. It is distinct from an outdated claim (which was once true) and from an entity gap (a fact that is missing rather than wrong).
See your AI Reality Score.
Find out whether AI describes your brand accurately, across all five engines.