The Citelligence Index

One score gets you in the door.
Two tell you what to do.

The Citelligence Index splits into two operator-readable scores: Defense (how well you're holding the ground you've earned) and Offense (how well you're taking new ground). Same six evidence-backed components, different prompt populations. Both want to go up. Refreshed every morning. Math is published.

DEFENSE
82
, Dominant ,
Holding 23 of 28 Core prompts · ▲ +5 · 30d
OFFENSE
41
, Emerging ,
Surfacing on 14 new prompts · ▲ +12 · 30d
Why two numbers

Dashboards show twenty metrics.
Operators need two.

One composite was gameable
  • ✗Add a losing aspirational prompt, score drops
  • ✗Hide aspirational prompts, score looks great, you're standing still
  • ✗Citation count up 14%, but where? On held ground or new ground?
  • ✗"Are we winning?" gets a 12-chart dashboard back
Two scores separate held from new
  • ✓Defense: how well you're holding ground you've already won
  • ✓Offense: how well you're taking ground you don't yet hold
  • ✓Same six components, different prompt populations
  • ✓Both want to go up, you can't fake one by hiding the other
One number is gameable on the denominator: add a losing aspirational prompt and your score drops, so the incentive becomes hiding losing ground. Two numbers separate "did you keep what you had" from "did you take new ground", both directional, both want to go up.
Scoring Bands

Know exactly where you stand.
And how far you have to go.

0 – 24
Invisible
AI doesn't know you exist. Competitors own every citation. Urgent attention required.
25 – 49
Emerging
You're surfacing on some queries, missing on others. Foundation is building but not yet defensible.
50 – 74
Competitive
You hold citations across most core queries. Now the work is defending position and expanding territory.
75 – 100
Dominant
You're the brand AI recommends first, across most platforms. The goal now is compounding the lead.
Bands are labels, not targets. What matters is the movement: a brand climbing from 34 to 48 has a better story than one holding steady at 72.
The Six Components

Every number has a reason.
Every reason has evidence.

Six weighted components. Same six run against your Core prompts (Defense) and your Conquest + Discovery prompts (Offense), same yardstick, different ground. Each one independently scored from live data, decomposed so you can see exactly what's driving each score, and exactly what moves it.

Weight · 30%

Topical Authority

How many queries you own.
Scoring weight

The single strongest predictor of AI visibility. AI models trust brands they see cited repeatedly across a cluster of related queries. A handful of wins on high-volume terms is less valuable than consistent citations across the full territory.

What it measures
The share of tracked queries where your brand appears in AI answers, weighted by query volume and topical clustering. A brand cited on 60 of 100 prompts scores dramatically higher than one cited on 30, even at similar positions.
What moves it
Shipping deep hub/cluster content that answers related questions. Expanding your tracked prompt portfolio. Winning back absent queries through targeted content + schema + entity work.
What a strong score looks like
Cited on 50%+ of tracked prompts. Winning on both head terms and conversational long-tail. Balanced coverage across hubs rather than lopsided into one cluster.
Weight · 25%

Entity Strength

How well AI knows who you are.
Scoring weight

AI models anchor recommendations to entities, not URLs. When the model can confidently answer "who is this brand?" it cites them with authority. When it can't, it defaults to the better-known competitor. Entity strength is your identity foothold in the AI knowledge graph.

What it measures
Schema.org Organization markup with sameAs edges. Knowledge-panel presence. Wikipedia / Wikidata entity existence. Citation alongside entity verifiers (LinkedIn, Crunchbase, Reddit, Medium). Name consistency across authoritative sources.
What moves it
Deploying Organization schema with sameAs links to Wikipedia, LinkedIn, Crunchbase, Reddit, Medium. Consistent brand naming across platforms. Off-site entity building on high-trust profiles that AI models treat as authoritative.
What a strong score looks like
Knowledge panel or Wikipedia entity present. Organization schema with 5+ sameAs edges. 90%+ name consistency across platforms. Cited alongside entity verifiers in AI responses.
Weight · 20%

Citation Density

How often AI mentions you, and how visibly.
Scoring weight

Position matters. Being named first in an AI answer carries dramatically more intent capture than being named fifth. Citation Density weights each mention by its rank within the response, rewarding the brands AI leads with.

What it measures
Every mention of your brand across the sweep, ranked by position in the AI response and frequency across queries. A brand named first on 20 queries outscores one named fifth on 40.
What moves it
Winning position #1 on core queries. Expanding the set of queries where you're cited at all. Owning the "top pick" framing through content that reads as the definitive answer, not one option among many.
What a strong score looks like
Position #1 on 30%+ of cited queries. Cited on every tracked core prompt. Named in the "top pick" framing rather than the "also available" list.
Weight · 10%

Structured Data

What you've deployed for AI to read.
Scoring weight

AI models grade sources by how machine-readable they are. Schema markup, sameAs links, and structured product data make it easier for the model to cite you with confidence. It's the difference between a page the AI has to parse and a page the AI can trust.

What it measures
Presence and correctness of Schema.org types across your pages: Organization, Product, FAQ, HowTo, Article, Review, plus 10+ additional types matched to page intent. Sameas edges to authority profiles. Breadcrumb and navigation schema.
What moves it
Deploying appropriate schema per page type. Adding sameAs markup to Organization schema. Product schema with aggregateRating and reviews. FAQ schema on answer-heavy pages.
What a strong score looks like
80%+ of pages carry appropriate schema. Organization schema with 5+ sameAs edges. Zero schema validation errors. Schema matches page intent.
Weight · 10%

Surface Coverage

How many AI platforms cite you.
Scoring weight

Being #1 on ChatGPT isn't enough. Your customers also use Perplexity, Gemini, Claude, Google AI Overviews, and DeepSeek, and each platform decides independently who to recommend. Surface Coverage rewards brands that are cited across the ecosystem, not just dominant on one engine.

What it measures
The number of AI platforms where your brand is cited, weighted by platform share. A brand cited on 5 of 6 platforms scores higher than one dominant on ChatGPT but absent everywhere else.
What moves it
Platform-specific optimization. Winning citations on under-represented platforms through entity building, schema, and content that matches each platform's training data preferences.
What a strong score looks like
Cited on 5+ of 6 tracked AI platforms. Balanced coverage rather than lopsided into one engine. No platform where you're completely absent.
Weight · 5%

Sentiment

How positively AI describes you.
Scoring weight

Being mentioned isn't winning, being mentioned favorably is. Sentiment captures the framing around your brand in AI answers: are you the "top choice," the "premium option," a "specialist", or just "also available"? Same citation, very different commercial outcome.

What it measures
Sentiment score of the sentence(s) containing each brand mention in AI responses. Classifier ranges from strongly negative (avoid, outdated) to strongly positive (top pick, definitive, leader).
What moves it
Winning third-party reviews and editorial content that AI models will train on. Shipping content that frames your brand in the exact language you want AI to echo. Addressing negative reviews that are shaping the sentiment floor.
What a strong score looks like
Consistently positive framing in AI answers ("top pick," "leader," "premium"). No negative or neutral "also available" mentions on core queries. Sentiment stable or improving week-over-week.
Evidence Trail

Every score traces back
to the evidence that made it.

The Index isn't a black box. Every component score is defended with the underlying data: the queries probed, the responses captured, the citations extracted, the sentiment classified. If a number moves, you can see exactly why.

What you get
  • ✓Full response text for every AI answer captured
  • ✓Position and context for every citation
  • ✓Competitor mentions with position data
  • ✓Sentiment classification per mention
  • ✓Component-level score decomposition on demand
  • ✓Week-over-week deltas with named drivers
What you don't get
  • ✗Vague "it went up" narratives
  • ✗Scores you can't explain to your board
  • ✗Proprietary weights hidden from view
  • ✗Claims that can't be defended with the data
Read the full methodology
Your Defense + Offense

See both numbers.
See the moves.

Start tracking your Defense and Offense scores this morning. 14-day free trial, no contract.

Start free trial Open sample dashboard