Marketing teams need numbers. “We’re doing AI optimisation” is not a number. “We moved our AI Visibility Score from 38 to 67 in 90 days” is. This article explains exactly what the AIOX Visibility Score measures, how it’s computed, and how to use it to track progress.
What the score measures
A single 0-100 number, refreshed daily, that quantifies how visible your brand is to LLMs across the four major providers (Gemini, ChatGPT, Claude, Perplexity). Higher = more visibility. The score is a weighted blend of five dimensions:
- Presence (30%) — does the LLM mention your brand by name when asked relevant questions?
- Citation (25%) — does it link to or quote your content directly?
- Accuracy (20%) — when it talks about you, is what it says correct?
- Sentiment (15%) — is the framing positive, neutral, or critical?
- Position (10%) — when listed among alternatives, are you mentioned first, second, or last?
How it’s computed
Every morning, AIOX runs a battery of probes against each provider. The probes are derived from the topics you’ve configured: your category (“CRM for small teams”), your products, your competitive comparisons. For each probe, the response is parsed for the five dimensions above.
The dimensions are normalised per provider (each LLM has slightly different baseline behaviours), then averaged across providers with equal weight (so no single vendor’s quirks dominate). The result is the daily Visibility Score.
Why daily, not real-time
LLM responses to the same prompt aren’t deterministic. Run the same query twice; you get two slightly different answers. A real-time score would jitter wildly. Aggregating across the daily probe batch (typically 20-50 probes per provider) smooths the noise and gives you a stable signal.
The 30-day rolling chart in the dashboard shows you the trend, not the spot value, which is the metric you actually care about.
What a “good” score looks like
Calibrating expectations:
- 0-25: LLMs don’t know about you. New brand, niche topic, or content not in any training corpus / retrieval index.
- 25-50: Sporadic awareness. LLMs mention you sometimes when asked directly but won’t surface you in comparison or “best X” queries.
- 50-75: Solid visibility. You appear in most direct queries and frequently in comparison queries.
- 75-90: Strong visibility. You’re a likely answer for category questions; competitors mention you when LLMs ask about them.
- 90+: Dominant. Your brand is the LLM’s default answer for your category. Rare.
What moves the score
From customer data:
- Bulk-processing your archive: +5 to +15 in the first 4 weeks. Capsule infrastructure makes retrieval-LLMs (Perplexity, ChatGPT Search) cite you more.
- Publishing a definitive piece on a topic you’re underrated for: +3 to +8 for that topic, with sitewide spillover.
- Fixing a stale-info issue surfaced by AI Forensics: +2 to +5 once the correction propagates.
- Adding cryptographic authorship: +1 to +3 (trust signals matter, on the margin).
- Earning third-party citations: +5 to +20 over 6 months. The strongest single lever.
What doesn’t move the score
- Keyword stuffing (LLMs don’t reward it the way Google sometimes does).
- Backlink count (helpful indirectly via the citation dimension, but raw count alone doesn’t move the dial).
- Page speed (matters for SEO; LLMs don’t directly weight it).
- Social-media metrics (LLMs don’t ingest most social platforms).
