Where AI sends your guests, city by city.

ChatGPT, Claude, Gemini, Perplexity and Grok, asked about every hotel you run. Every answer scored.

Which hotel would AI recommend?

Location reportMeridian BrickellBrickell, Miami
61Fair
Meridian Hotels Group37 hotels in 6 pilot cities, demo portfolio
  • 80+ Excellent
  • 60-79 Fair
  • Below 60 Needs work
  • Target 70
  • Pulsing hotels open a report

Ask like a guest

in Miami

One question is one sample. We ask 100 per hotel.

Each question is about the hotel lit on the map. Demo answers.

ChatGPT

Claude

Gemini

Perplexity

Grok

The five assistants your customers already ask

  • ChatGPT
  • Claude
  • Gemini
  • Perplexity
  • Grok

Guest reviews, from where guests leave them.

Reviews from Tripadvisor, Google Maps and Booking.com sit beside the AI answers for every location, the good ones and the bad.

Sample reviews of fictional businesses.

How a score is made.

One hotel, from its website to its first fix: Meridian Brickell in Miami, start to finish.

  1. Youname and websiteWeresearch

    Give us the brand's name and website. We read the site, guest reviews and press, then write a brief you can edit. Every question starts from it.

  2. Youapprove

    Hotels get ten categories with ten topics in each: 100 questions for every hotel, written from its own facts and address. You approve the categories before anything runs.

  3. Weask

    ChatGPT, Claude, Gemini, Perplexity and Grok each answer all 100. For one hotel that is 500 answers, kept word for word.

  4. Wescore

    Each answer is read for whether the hotel is named, recommended, and at what rank, then scored from 0 to 100. The hotel's score is the mean, with its spread beside it. Meridian Brickell lands at 61, under the target of 70.

  5. Youact, with an owner named

    The answers become a short list of decisions. Each one names an owner and opens to the answers that raised it. Here, three assistants still tell guests the rooftop pool is closed.

  6. Wetrack

    Every run is compared with the last one. A move under 3 points is marked as noise, so a real change stands out, and each person gets a digest on their own schedule.

Same brand. A different answer in every city.

Every location is asked the same questions by all five assistants. They rarely agree, and the gaps are where customers end up somewhere else.

Meridian Gateway, Newark. Recommended first by one of five assistants. Perplexity warns guests off it and points them to a competitor.

Hotel ChatGPT Claude Gemini Perplexity Grok Score

Demo data. Each cell is one assistant's score for one hotel.

The whole brand on one map. Every location in one report.

1

Brand map

Every location on one map, colored by score, with the portfolio's average, best and worst beside it.

The brand map: scored locations across Miami, the overview card and the list of locations
2

Location report

The score with its spread, sentiment, each assistant's score, the first thing to fix and the ten categories.

A location report: the score with its spread, sentiment, assistant scores, the first decision and ten categories

Four things it does for you, in motion.

Find every location

Search a city and we list your properties from Google. Add them to the map in one step.

A brief researched for you

We read the brand's website, reviews and press, then write the brief the questions are built on.

A score with its spread

Each score shows where most answers landed, so a small move between runs is not mistaken for a trend.

A digest for each person

Every teammate gets the locations they follow, on the schedule they choose.

Measure, explain, act.

Measure

  • A score for every locationFrom 0 to 100, banded Excellent, Fair or Needs work, against a target of 70.
  • Five assistants side by sideWhere they agree, and where one of them is off.
  • Categories for your industryTen topics in each, approved by you.

Explain

  • The answers behind every numberQuoted word for word, with the question that was asked.
  • Sources, namedThe sites each assistant cites for what it says.
  • Competitors named insteadWho takes the answer when you are left out.

Act

  • Decisions with an ownerWhat to change first, and who does it.
  • A read-only share linkThe map and reports for people without an account.
  • Email digest and PDFA schedule for each person, and a file for the board.

See your locations the way AI assistants describe them.

We run the analysis on your own locations, then walk you through what the answers say.

Get early access

Fill in your details and we will get you set up with early access to Geo Percept.