Overview

The Goodie AI Visibility Index measures how brands, products, and sources appear inside AI-generated answers, not traditional search rankings, but what large language models actually say, cite, and recommend when someone asks a real question in their category.

Every figure on this site, a visibility score, a share of voice, a cited-domain ranking, is rebuilt on a fixed monthly cycle from the same methodology, so a number from one month can be compared directly against the next.

Real intent capture

The index starts from real user prompts, not brand-supplied keywords. Each tracked industry is represented by a set of prompts modelled on how people actually ask about it: comparisons, recommendations, "best of" questions, and troubleshooting. Prompts are captured and categorised with semantic and agentic processing that filters out noise and keeps only high-value, on-topic queries.

Global model interrogation

Every prompt in an industry's set is run against every tracked AI model, currently AI Mode, ChatGPT, Copilot, Perplexity, across every tracked region, on a monthly cadence. Running the same prompt across models and regions is what lets the index show where models agree, where they diverge, and how a market changes the answer.

Scoring and attribution

A visibility score reflects how often and how prominently a brand appears across sampled answers for its industry; share of voice expresses that as a proportion of the whole category rather than an absolute number. Separately, the index analyses every generated answer for the sources models cite, the domains that ground a category's answers, to map which sites are actually driving citations, and how much of that citation share belongs to a brand's own domain versus third parties.

Publish tiers

An industry only publishes a full ranking once it has collected enough monthly observations to be statistically stable. Until then it is shown as a preview with partial data, or held back entirely. The index does not guess at a ranking it does not yet have the sample size to support.

Limitations

Coverage is limited to the industries, models, and regions currently tracked, and a monthly cadence means the index reflects a rolling window rather than real time. Models are non-deterministic and individual answers can vary between runs; the index reports patterns observed across many sampled runs, not a single query.