There's no 'accuracy' in AI search tracking — only precision, argues Mike King
VeryWellVersed · x · 2026-09-18
When two AI visibility platforms report 41% vs 23% brand visibility, asking which is 'accurate' is a category error, argues Mike King of iPullRank. LLM outputs are probabilistic distributions with no true value to compare against — tools can only offer precision. Methodological choices (scraping vs API, logged-in vs out, sampling frequency) are the measurement, and most buyers never learn how their tool samples. His advice: scrutinize each vendor's methodology before making expensive decisions.
More from Apps
- Perplexity rolls out effort controls in Computer; user shares a 3-tier model workflow — inductionheads · 2026-09-18
- Google Research uses generative UI to let teachers create learning interactives — gaganghotra_ · 2026-09-18
- AI CFO startup cfo.ai says Ari is growing faster than any month in its history — soleio · 2026-09-18
- Lapis launches: a low-effort site for publishing what matters to you — almmaasoglu · 2026-09-18
- Google Research uses generative UI to let teachers build learning interactives without code — Google Research · 2026-09-18
- Pika launches new AI creative platform aggregating Seedance, Veo, GPT Image and more — LudovicCreator · 2026-09-18