When we built the AI Clarity Index, we started with a simple question: how do you measure something that didn’t have a measurement system?
Businesses have had tools to measure search rankings for two decades. Google Analytics, Ahrefs, SEMrush — the infrastructure for understanding your Google presence is mature. But for AI visibility — what ChatGPT, Gemini, Claude, and Perplexity say about your business — there was nothing.
No score. No dashboard. No way to know if what AI tells your potential customers is accurate, incomplete, or completely wrong.
Here’s how we built one.
The Process: 20 Questions Across 4 Models
Every AI Clarity Index audit starts the same way. We query four major AI language models — ChatGPT, Gemini, Claude, and Perplexity — with 20 structured questions about your business.
The questions fall into five categories:
Direct queries. “Tell me about [business name].” These test whether AI recognizes your business at all and what it knows about you.
Category queries. “What are the best [your service] providers in [your city]?” These test whether AI includes you in relevant category recommendations.
Recommendation queries. “I need a [your service]. Who should I use in [your area]?” These test whether AI actively recommends you when asked.
Comparison queries. “Compare [your business] and [competitor].” These test whether AI differentiates you accurately from competitors.
Authority queries. “Is [your business] reputable? What do people say about them?” These test AI’s assessment of your credibility and reputation.
Each model answers each question independently. That gives us 80 data points per audit — 20 questions times 4 models — which form the basis for scoring.
The Six Metrics
We score every audit across six proprietary metrics. The first three measure per-model accuracy. The last three measure cross-model patterns.
Per-Model Metrics
Entity Clarity Score (ECS). Does AI correctly identify your business name, location, and core identity? A low ECS means AI confuses you with another business, misidentifies your location, or doesn’t recognize you as a distinct entity. This is the most fundamental metric — everything else depends on AI knowing who you are.
Source Alignment Score (SAS). Are AI-generated claims about your services, history, and offerings factually accurate? SAS measures the gap between what AI says and what’s actually true. A business might have high Entity Clarity (AI knows who you are) but low Source Alignment (AI gets your services wrong).
Narrative Consistency Score (NCS). Does a single AI model tell a consistent story about your business across different question types? Sometimes a model will describe your business accurately in a direct query but contradict itself when answering a recommendation or comparison question. NCS catches these inconsistencies.
Ensemble Metrics
Cross-Model Variance (CMV). How wide is the gap between the most and least accurate models? If ChatGPT gives you an 85 but Perplexity gives you a 40, your customer experience depends entirely on which AI they happen to use. Low variance means consistent representation regardless of model.
Consensus Confidence Index (CCI). When models agree about your business, how confident and correct is that agreement? High CCI means multiple models say the same accurate things about you. Low CCI means models agree — but they’re agreeing on something wrong. This is a subtle but important distinction.
Drift Risk Indicator (DRI). Are there signals that your AI visibility will degrade over time? DRI looks for warning signs: reliance on a single outdated source, thin content signals, rapidly changing competitive landscape, or inconsistencies that suggest models are working with stale data. A high DRI means your current score may not hold.
The Composite Score
The six metrics combine into a single ACI Score on a 0–100 scale, weighted to reflect their relative importance:
- Entity Clarity (ECS): 20%
- Source Alignment (SAS): 20%
- Narrative Consistency (NCS): 15%
- Cross-Model Variance (CMV): 15%
- Consensus Confidence (CCI): 15%
- Drift Risk (DRI): 15%
ECS and SAS carry the most weight because they represent the most fundamental aspects of AI visibility. If AI doesn’t know who you are or gets your services wrong, the other metrics are secondary.
Score tiers:
- 90–100: Excellent. AI represents your business accurately and consistently across models.
- 70–89: Good. Core identity is solid with some gaps to address.
- 50–69: Fair. Significant inaccuracies or inconsistencies that affect customer perception.
- 0–49: Poor. AI either doesn’t know your business or actively misrepresents it.
What the Report Includes
The free preview report shows your composite ACI Score and three metric labels — enough to understand where you stand.
The full audit adds the complete picture: individual scores for all six metrics, per-model breakdowns showing exactly what each AI says, specific findings with evidence (direct quotes from AI responses), and prioritized recommendations ranked by impact and effort.
The Audit + Strategy tier includes competitor comparison, a 60-minute strategy session, and a 90-day re-audit to measure progress.
Why These Metrics Matter
Every metric in the AI Clarity Index maps to a specific business risk.
Low Entity Clarity means potential customers asking AI about you might get directed to a competitor. Low Source Alignment means AI is telling people the wrong things about your services. High Cross-Model Variance means your reputation depends on which AI app someone happens to open.
These aren’t abstract scores. They represent real scenarios where potential customers form opinions about your business based on what AI tells them — before they ever visit your website or read a review.
Check Your Score
The free preview report takes two minutes. Enter your business details and we’ll query all four models, score the results, and show you where you stand.
Get Your Free Preview Report →
Most businesses score between 45 and 70. The ones that score higher aren’t necessarily bigger or better known — they just have stronger entity signals across the web. That’s fixable, and the first step is knowing your number.