LLM Visibility Score
See how five large language models rate your brand - one 0-100 score, three levers, and the exact fix that moves it most. Free.
What is an LLM visibility score?
An LLM visibility score measures how consistently large language models recommend your brand when buyers ask about your category. Unlike traditional search rank, LLMs source answers from both training data and live web retrieval - a brand can rank page one on Google yet score near zero here, and vice versa. CitedSpy finds your competitive set, picks the real buyer prompts where you should appear, queries ChatGPT, Perplexity, Gemini, Claude and Grok live, and returns one 0-100 score with a three-lever breakdown.
Large language models learn about your brand through two distinct channels. The first is training data - the text ingested before the model's knowledge cutoff. If your brand is discussed across credible sources, comparison posts, directories, and published reviews, the model develops an internal representation of who you are and what you do, independent of whether it can search the web. The second channel is live retrieval: ChatGPT with web search, Perplexity, and search-augmented Gemini issue real-time queries and synthesise fresh results before answering. A brand absent from training data but well-cited on the live web can still score high on those engines - while a brand strong in training data but under-cited online will disappear the moment an engine switches to web-search mode.
An LLM visibility score measures how you perform across both channels together. It is not a ranking signal - Google's PageRank and LLM recommendation are powered by different machinery. A 2026 Ahrefs analysis of 863,000 keywords found that only about 38% of AI Overview citations come from pages ranking in Google's top 10, with the remainder split roughly evenly between positions 11-100 and beyond position 100. A well-reviewed brand known across authoritative sources can score high on an LLM query even without first-page SEO. Your LLM visibility score is a direct measure of how AI-native buyers discover your brand - separate from, and often misaligned with, your traditional search rank.
CitedSpy scores LLM visibility across three levers that add up to one number. Presence is the share of real buyer prompts where at least one LLM names you - the biggest lever, because simply showing up is most of the battle. Rank quality captures where you land when you do appear: being recommended first is worth far more than fifth, since LLM answers typically lead with the strongest recommendation. Sentiment measures how positively models describe you, scaled by presence so a brand that never appears cannot inflate its score on tone. The breakdown shows not just your number but which specific lever to pull first.
Score your LLM presence in four steps
Enter your brand name
Type your brand as buyers search for it. We identify your category and the real prompts where LLMs should recommend you - so the score reflects your actual market, not a guess from the name alone.
Add your domain (recommended)
With your domain we fetch your homepage directly, which is what makes niche and local brands work. It also lets us tell when an LLM cites your own site - a sign of deep recognition, not just a passing mention.
We query five LLMs live
CitedSpy finds the ~5 competitors LLMs are most likely to recommend in your space, picks the real buyer questions where you should rank near the top, and queries ChatGPT, Perplexity, Gemini, Claude and Grok live across each prompt.
Read your LLM visibility score
Get a 0-100 score with the three levers behind it - presence, rank quality and sentiment - a per-LLM spread, and the single fix that would lift your score most.
How to read your LLM score
Start with the headline. Below about 15 your brand is largely invisible to large language models in your own category - they recommend competitors instead. Around 40-70 is partial visibility: LLMs know you and mention you sometimes, but rivals crowd you out on several engines or prompts. Above 70 is strong - LLMs reliably surface you near the top of category answers. But the band alone does not tell you why, which is what the breakdown is for.
The per-LLM spread is where the real intelligence sits. It is normal to score well on Perplexity and ChatGPT with search (which lean on live web content) while scoring lower on an engine working from older training data - or vice versa. A strong score on training-data-heavy LLMs signals your brand is embedded in the authoritative sources those models ingested. A strong score on live-retrieval LLMs means you earn citations from fresh, high-authority content. Knowing which engines you win and which you lose narrows the fix: more citation-worthy content for live-retrieval gaps, or deeper third-party coverage for training-data gaps.
What to remember
- An LLM visibility score is a single 0-100 number for how consistently large language models recommend your brand in category queries.
- LLMs learn from two sources - training data (past knowledge) and live web retrieval (present search) - so LLM visibility and Google rank are separate and often misaligned measures.
- Presence is the biggest lever: showing up in LLM answers at all matters more than ranking high or being described positively.
- The per-LLM spread reveals which engines know you and which do not - the right fix is different for training-data gaps versus live-retrieval gaps.
- LLM answers drift as models retrain and live sources update, so a one-off score goes stale fast and needs to be tracked over time.
Our methodology
A score runs in two phases. First, CitedSpy works out who you are: when you provide a domain we fetch and read your homepage directly, then use that plus live web search to identify your category, the roughly five competitors LLMs tend to recommend in your space, and a set of natural buyer questions where you should rank near the top. Reading your own site is what makes the score meaningful for niche and local brands that language models may not have encountered in training data.
Second, it sends every prompt to each available LLM - ChatGPT, Perplexity, Gemini, Claude and Grok - using their live web-search tools where supported, and reads each answer for whether you are named, in what order, and how positively. From that it computes three sub-scores - presence, rank quality and sentiment - and combines them with fixed weights of 60, 25 and 15 into a single 0-100 figure. Sentiment is multiplied by presence, so tone can never lift a brand the models do not mention.
The score rewards genuinely showing up and ranking high over a bare name-drop. Only LLMs actually reached are counted - a slow or rate-limited LLM is reported as not reached, so it can never silently deflate your number. Results are cached for 24 hours so a repeat check is instant. This is the same measurement engine behind CitedSpy's paid tracking: the free tool scores you once, the product re-scores on a schedule and shows the trend.
LLM Visibility Score, answered
Related tools & reading
Stop guessing. Track your AI visibility.
Run this across all engines on a schedule, watch competitors, and get alerted when your brand slips from AI answers.