The Temso Methodology

What we measure

Rankings are derived from Temso's AI Visibility Index, which measures how brands, publishers, and social channels appear in answers returned by the AI search engines people actually use: ChatGPT, Google AI Overviews, Microsoft Copilot Search, and Grok Search.

How we collect data

For each industry, category, and target country, Temso runs a fixed set of localized prompts every day for the full calendar month. Queries are issued through the real consumer-facing interface of each AI search engine, routed through residential proxies in the relevant country so that responses reflect what an actual user in that market would see, not an API surface or sanitized endpoint. The current rankings are computed from over 500,000 search results captured across the four engines during the period.

Signals we extract

Two signals are extracted from each captured response:

• Brand mentions, identified through entity matching against the response text and deduplicated per response.
• Cited sources, parsed from the URLs each engine returns as references and resolved against a curated domain registry that classifies every domain as a publisher (editorial) or a social channel (user-generated content).

Metrics

Two metrics are computed from those signals.

Visibility (for brands): the share of unique responses in a segment where the brand is mentioned, computed as the number of unique responses mentioning the brand divided by the total responses in the segment.

Presence (for publishers and social channels): the share of unique responses in a segment where the source is cited, computed as the number of unique responses citing the source divided by the total responses in the segment.

Segmentation and ranking

Brands are ranked within each industry, category, and country. Publishers and social channels are ranked at both the category and industry level, and an aggregated Global view is produced alongside the country-specific segments. Each row carries a change value equal to the absolute difference between the current month's score and the prior month's score in the same segment.

Cadence

Rankings are recomputed on the 1st of each month over the previous full calendar month and stored as immutable monthly snapshots, so movement between snapshots reflects true period-over-period shifts in AI visibility across the four search engines.