AuthoritySignal Research analyzed 480 completed observations across 30 anonymized active New York real estate salespersons, Perplexity sonar, Gemini gemini-2.5-flash, and eight fixed consumer-intent query families. The collection ran August 27, 2026, from 00:02:52 to 00:27:05 UTC.
No observation was classified as recommended or recommended prominently: 0 of 480 observations and 0 of 30 sampled subjects. This descriptive result applies only to the sampled subjects, providers, query families, and collection window.
Recommendation visibility was 0 of 240 completed observations for Perplexity and 0 of 240 for Gemini. Each of the eight query families was 0 of 60 across both providers.
Entity recognition was resolved in 0 of 480 usable observations; 27 were ambiguous and 453 were not resolved. Perplexity returned one or more source references in 240 of 240 responses. The Gemini integration did not expose comparable citation data, which is not treated as zero citations.
Recommendation visibility counts completed observations classified as recommended prominently or recommended, divided by usable completed classifications. All 480 planned observation cells completed; there were no unknown, blocked, timed-out, malformed, or provider-error observations.
This study is descriptive, not causal or nationally representative. It does not test factual accuracy, market differences, competitive concentration, or which signals influence recommendations. Public output excludes identities, exact prompts, raw responses, and restricted evidence references.