What biases distort AI visibility measurement?

Prompt-driven bias comes from an unbalanced query set. A set weighted toward recommendation prompts flatters brands that do well in recommendation contexts and understates them everywhere else. The control is disclosing the prompt taxonomy so a reader can judge whether the set is representative.

Platform-driven bias is different: one model’s training data or response behavior systematically favors or penalizes you, independently of your actual market position. The control is never blending platforms into one score.

Both are why a single headline number is the least trustworthy thing a visibility tool can show you. Run identical query sets across platforms and read the per-platform results.