We audited 93 companies and logged 39,046 source links behind AI answers. Only 1.5% pointed at the company itself. How ChatGPT builds a recommendation, why you can't pay for one, and what earns a place in it.
We ran our own AI-visibility tool on our own site: 351 AI answers across five engines, zero cited citepulse.ai. Why AI citations can't be bought, what a 45-study GEO survey shows, and what earns them instead.
Every plan now measures weekly. You add your own buying questions on top of the audit series. AI bot logs show which pages GPTBot, ClaudeBot and PerplexityBot read, crossed with what AI cites. Draft Studio assembles FAQ, comparison and freshness blocks from your approved facts, with no language model. And why Copilot shows as “not measured”.
Run the same buying question on consecutive days and about two thirds of the cited sources change. Why a before/after needs a noise band, how CitePulse measures it on a public daily panel (NOISE-1.0), and what “seen in k of n runs” means.
51% of B2B buyers now start research inside an AI, not Google, and 85.7% of AI citations point to third-party sites, not your domain. The research behind AI invisibility, the dark funnel, and the four layers that decide whether AI cites you.
A small brand beat three fashion giants 29 mentions to 2 in AI search, and Anthropic just shipped a browser tool that reads structure, not looks. The same rule decides both: AI rewards specific and legible, not big.
Google's AI Overview does not rank your page, it extracts a sentence and names the source. What "built to be extracted" means, four moves that make a section quotable, and how to know if it works.
AI answers are built from self-contained passages, not whole pages. What Google's patents suggest about extraction, why a page can rank and never be quoted, and how to make your content extractable.
How to check whether Perplexity cites your brand, the visibility metrics that matter (citation frequency, share, mention vs citation, source presence), how to improve them, and how to monitor Perplexity visibility over time.
Rankings are decoupling from being chosen. Citation share (whether AI engines cite your domain on real buyer questions) is the metric that replaces impressions, clicks and rankings. The evidence (Ahrefs, Pew, Similarweb, arXiv) and a 25-company benchmark with a median AI visibility of 12/100.
The six technical blockers that keep businesses out of AI answers (robots.txt, JavaScript-only rendering, missing schema, thin claims, no third-party citations, category ambiguity) plus a free step-by-step diagnosis and honest timelines for each fix.
97% of llms.txt files never get fetched and no study shows a citation effect. What the file is, who reads it, why CitePulse detects it but refuses to score it, and the one case where it helps.
A simple manual method to check whether AI assistants name your brand today, why one-off checks aren't enough, and how CitePulse.ai turns it into a repeatable measurement.
A repeatable, multi-engine method: define real buyer questions, run them on every engine, and repeat on a cadence with frozen prompts so you get a trend, not a screenshot.
What GEO is, how it differs from SEO and AEO, the metrics that matter, and how to track whether AI cites and recommends your brand across ChatGPT, Perplexity, Gemini and Claude.
An honest, sourced side-by-side of CitePulse, Otterly.AI, Profound, Peec AI and Knowatoa: free access, AI engines covered, pricing and agency white-label. Including where each competitor beats us.
Search is moving from blue links to direct answers. We explain what AEO is, why most companies are now invisible to ChatGPT, Perplexity and Gemini, and the real data behind the shift.
A concrete checklist for making your site readable and citable by AI crawlers (GPTBot, PerplexityBot, Google-Extended) mapped to exactly what a CitePulse audit verifies.
Citation frequency, source presence, and competitor displacement: the factors that determine whether an AI assistant names your company, and how CitePulse measures each one.