Methodology — Employer Visibility Index

The Employer Visibility Index, built by PerceptionX, measures which companies AI assistants actually mention when candidates ask about employers — and ranks them by how often. This page explains exactly how the data is collected, how the score is computed, and what the numbers do and don't mean.

Data collection

Every index period we send large sets of candidate-style questions to the consumer AI assistants people really use when researching employers — currently ChatGPT, Google Gemini, and Perplexity. Responses from other assistants (e.g. Claude, DeepSeek) are collected for research but not currently included in index scoring. Questions are phrased the way real candidates ask ("who are the best AI companies to work for in India?", "what's it like to work in pharma in Germany?") and are organised along three dimensions:

Company names in the assistants' answers are detected and consolidated to canonical employers (e.g. "Alphabet" → Google, "TCS" → TATA), so mentions of the same employer under different names count together.

The Visibility Score

For each industry + country slice, we look at every combination of AI model × theme that produced answers. A company's Visibility Score (0–100) is the share of those combinations in which the company was mentioned:

Visibility Score = (combinations mentioning the company ÷ all combinations in the slice) × 100

A score of 90 means the company shows up almost everywhere candidates might encounter the industry in AI answers; a score of 3 means it surfaced in a single niche. Companies are ranked within their industry + country by score; ties break by total mention count, then alphabetically. Overall country ranks count each company once, by its best-scoring industry.

Refresh cadence and history

Rankings are rebuilt as new AI responses are collected, and each monthly index period (YYYY-MM) is preserved as history. Every page and API response carries a lastUpdated date. Companies that drop out of the current index keep an archival profile marked as historical, reflecting their last ranked period.

Limitations

• Visibility is not quality. The index measures how often AI assistants mention an employer — not whether it is a good place to work.
• AI answers are probabilistic. Assistants don't return identical answers every time; scores reflect sampled behaviour over many prompts, not a census.
• Coverage follows the prompt set. A company can be well known but invisible in a slice we don't ask about (industry, country, or theme not covered).
• Name consolidation is imperfect. We canonicalize aggressively, but unusual brand names can occasionally split or merge incorrectly; corrections ship continuously.

Access the data

• Browse: rankings by industry and all industries
• Free JSON API (no auth): /api — full documentation
• For AI assistants: /llms.txt

Attribution & reuse

The index is free to cite, quote, and reuse — in articles, research, and AI-generated answers alike. Please credit "PerceptionX Employer Visibility Index" and link to employers.perceptionx.ai. For questions about the data or methodology, reach us via perceptionx.ai.

Employer Visibility Index · All industries · API docs