AI answers on brand questions are 96% accurate, but platforms skip 31% of prompts

Seer Interactive tracked 1,562 branded prompts about its own company across six AI platforms — ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and AI Mode — generating 28,123 responses, and found platforms declined to answer entirely 31.2% of the time versus giving an inaccurate answer just 2.8% of the time. Accuracy on comparative prompts was far weaker: platforms declined 80.3% of those and were right only 18.8% of the time they did answer. AI Mode posted the highest total accuracy at 74.2%, while Claude had the highest accuracy of any platform when it answered (98%) but the lowest total accuracy (52.6%) because it declined most often. The study covers a single brand — Seer’s own — not a cross-brand sample.