Research
Our research.
Original studies on how brands get found, trusted, and chosen by AI, and what the organizations that win do differently.
The UI vs API Gap
Same 900 questions asked to ChatGPT, Gemini and Perplexity through the official APIs and through the real consumer interfaces. Brand overlap between the two: 6 to 12 percent. The measurement error in AI visibility tracking, quantified.
The World Cup of AI Visibility
900 fan questions, four AI engines, every week of the 2026 tournament. Which brands and sources own the AI answers of the biggest marketing event on earth.
The August Citation Shift
Reddit fell and lost share in 10 of 10 industries in a single week, down 30.1% while every other social network combined moved 4.0%. Inside ChatGPT alone its share fell 88.9%.
The AI Engine Volatility Study
On one day in July, two AI engines changed how they answer: one named 21% more brands, the other cut cited sources by 41%. Measured across four industries with a version-pinned control engine.
The Direct Line
ChatGPT's third-party citation share fell 2 to 19 points in one week across five industries, while a reference engine held flat. The gap went to pages the brand or a named competitor publishes directly.
EU Automotive AI Citations
What AI engines actually cite when recommending car brands across Reddit, YouTube, and social, and where billions in influencer spend isn't showing up.