why do i score high on one engine and low on another?

Updated June 30, 2026 · Reviewed by Mack Grenfell · Asked 48 times

Because each engine learns and answers differently. A pure training-data model like Claude knows you only if you appeared in quality sources before its cutoff date. A live-retrieval model like Perplexity searches the web first, so it reflects what is being published about you right now. Those are different mechanics, so the scores diverge, and that is a feature, not an error.

What a split tells you

  • **Strong on training-data models, weak on retrieval:** you have historical presence but are missing from the fresh sources models pull today. Work on citations and earned media.
  • **Strong on retrieval, weak on training-data models:** you are getting cited now but were not well known when older models trained. This usually improves slowly as new model versions ship.

How fast each moves

Live-retrieval engines (Perplexity, AI Overviews, ChatGPT Search) respond quickly to new content, so changes show there first. Pure training-data models move slowly and only really shift across model versions. When you make a change, re-measure the next day on the retrieval engines and be patient with the rest.

Related questions

Guides and further reading