As of September 16, 2026.

This is the immutable September 2026 edition — its numbers are frozen as published and won't change. See the latest edition.

124
Destinations tracked
79,884
Traveler questions asked
108,924
Places identified
4
AI assistants observed (ChatGPT, Claude, Gemini, and Perplexity)

Observing since July 23, 2026 (~8 weeks).

Six benchmarks

17%
Model agreement

AI assistants agreed on the same top recommendation in 17% of 7,167 comparable traveler questions, as of September 16, 2026.

Quorum-tolerant measure — a question counts once at least 3 of 4 assistants gave a clear top pick, across 124 destinations. Full breakdown.

80.6%
Confirmed-open rate

80.6% of places AI recommends are confirmed still open as of our latest check.

79%
Multiple names — among places that qualified

Among the ~15,290 places observed often enough to check consistently (10+ independent observations), 79% are referred to by more than one name across AI answers.

Scoped to places with enough observations to check consistently — not the full 108,924-place network. How we track this.

95%
Positive statements

AI travel advice is overwhelmingly positive — under 2.9 in 1,000 statements about a place are negative.

970
Permanently closed, still recommended
462
Temporarily closed, still recommended

AI continues to recommend 970 permanently closed and 462 temporarily closed businesses. Full report.

11.4%
tripadvisor.com — top citation share

tripadvisor.com accounts for 11.4% of citations — more than Reddit, Expedia, and Michelin Guide combined.

Based on Perplexity citations — currently the only assistant that exposes its sources to us, last 90 days of data.

What's changing

Naming

Among the places TakeScout has observed enough to check consistently — 15,290 of the full 108,924-place network — 79% are called more than one thing across different AI answers. This reflects places we've observed often enough to check, not the whole network; places with only a handful of observations aren't included because there isn't enough signal yet to tell a real naming inconsistency from a one-off phrasing.

Sentiment

95% of the statements AI assistants make about a place are positive, and well under 1% are negative. AI travel advice essentially doesn't voice criticism — a place with real, well-known drawbacks still gets described the same way as one without.

Sources

tripadvisor.com leads every other site AI's search draws on for travel answers, at 11.4% of citations. This is based on Perplexity citations only — currently the only assistant that exposes its sources to us — over the last 90 days of data.

What AI agrees about

Tracked destinations with the highest cross-model agreement (at least 4 comparable questions each).

  • Seattle 38% agree

    The assistants landed on the same top pick in 23 of 60 comparable questions.

  • Ottawa 37% agree

    The assistants landed on the same top pick in 22 of 60 comparable questions.

  • Toronto 34% agree

    The assistants landed on the same top pick in 20 of 59 comparable questions.

  • San Diego 33% agree

    The assistants landed on the same top pick in 20 of 60 comparable questions.

  • Atlanta 33% agree

    The assistants landed on the same top pick in 20 of 60 comparable questions.

  • Austin 32% agree

    The assistants landed on the same top pick in 19 of 60 comparable questions.

  • The assistants landed on the same top pick in 19 of 60 comparable questions.

  • Edmonton 30% agree

    The assistants landed on the same top pick in 18 of 60 comparable questions.

What AI disagrees about

Tracked destinations with the lowest cross-model agreement (at least 4 comparable questions each).

  • Abbotsford 2% agree

    The assistants landed on the same top pick in 1 of 45 comparable questions.

  • Wisconsin 3% agree

    The assistants landed on the same top pick in 2 of 60 comparable questions.

  • Saskatchewan 3% agree

    The assistants landed on the same top pick in 2 of 60 comparable questions.

  • Sherbrooke 4% agree

    The assistants landed on the same top pick in 2 of 54 comparable questions.

  • Nevada 5% agree

    The assistants landed on the same top pick in 3 of 60 comparable questions.

  • The assistants landed on the same top pick in 3 of 60 comparable questions.

  • New Brunswick 5% agree

    The assistants landed on the same top pick in 3 of 58 comparable questions.

  • The assistants landed on the same top pick in 3 of 56 comparable questions.

What sources AI rely on

Based on Perplexity citations only — the only assistant that currently exposes sources — last 90 days of data. Full breakdown.

  1. tripadvisor.com ↗ 12,947 citations
  2. reddit.com ↗ 3,252 citations
  3. expedia.com ↗ 1,696 citations
  4. guide.michelin.com ↗ 1,432 citations
  5. no.tripadvisor.com ↗ 1,369 citations
  6. ar.tripadvisor.com ↗ 1,349 citations
  7. nps.gov ↗ 1,348 citations
  8. booking.com ↗ 1,296 citations
  9. onlyinyourstate.com ↗ 1,091 citations
  10. opentable.com ↗ 1,035 citations

Most visible destinations

Ranked by composite score — the blend of agreement, confirmed-open rate, distinctiveness, and presence. How the score works.

Biggest movers

Trend data is still accumulating — biggest movers will appear starting with the next edition.

Methodology

Every number on this page is computed directly from TakeScout's tracked AI observations. Full methodology.