AI Simulation

The GEO score tells you how citable a page is. The AI simulation tells you what's actually happening — by asking real AI engines real questions and measuring whether they mention and cite you.

What it does

Citera sends a set of generated queries to five AI engines in parallel — ChatGPT, Claude, Gemini, DeepSeek and Perplexity — and analyses every answer for two things: whether your brand is named, and whether your domain is referenced as a source. The result is a real-world visibility snapshot across the engines your audience actually uses.

Query generation

Queries are generated from your page topic, keywords and brand, then grouped into categories:

• Discovery — generic questions a potential customer would ask without knowing your brand.
• Comparison — "best tool for…" or "X vs Y" style questions where alternatives compete.
• Brand-aware — questions that name your brand directly.

Your headline mention rate is measured on discovery and comparison queries only. Brand-aware queries name you on purpose, so including them would inflate the number — they're shown separately.

Mention vs. citation

These are two different signals, and Citera tracks them separately:

• Mention — your brand name appears anywhere in the AI's answer. Good for awareness.
• Citation — your own domain is referenced as a source. This is the stronger signal: the engine is pointing users to you.

A page can be mentioned without being cited, and vice versa. The goal of GEO is to grow both — especially citations on discovery queries where you're competing on merit.

Reading the results

Results are shown as a mention rate and citation rate, with a per-engine breakdown and a per-query view. A green check next to an engine means it mentioned you for that query; the number beside it is your position in the answer. Queries where no engine mentioned you — typically generic discovery questions — are your clearest opportunities to improve.

Competitor visibility

The same answers reveal who you're up against. Instead of guessing, Citera counts the domains the engines actually recommended across every response and ranks your competitors by real frequency. You see three views:

• Share of voice — your brand next to your top rivals, by how often each is recommended across all answers.
• Ranked competitors — each rival with its overall visibility and a per-engine count, so you can tell whether a competitor dominates everywhere or just on one engine.
• Breakdown by AI engine — each engine analysed on its own: your visibility there versus the rivals it recommends most. Engines disagree, and this is where you see it.

Rankings come from all five engines equally. Generic domains (search engines, social networks, encyclopaedias) and your own domain are filtered out so the list stays focused on real alternatives.

Sentiment & citation likelihood

Beyond "were you mentioned?", every simulation adds two deeper signals:

• Sentiment — for each answer that mentions you, Citera classifies the tone toward your brand as positive, neutral or negative. The Brand Visibility card shows a tone summary, and each engine pill is colour-coded so you can spot answers that mention you unfavourably.
• Citation likelihood — a 0–100 score per query estimating how strongly the engines are pointing at you, built from mention, position, whether your domain is cited, tone and your GEO score. Use it to prioritise: queries near the top are "almost there — optimize", low ones are "skip for now". It's a relative strength score, not a literal probability.