Wayfinder
Measuring and improving how AI assistants like ChatGPT, Gemini, Google’s AI Overviews and Claude identify and represent rural businesses.
Why it matters
Travelers increasingly plan trips by asking an AI assistant a question and acting on the answer they get back. If the assistant can’t read a business, or reads it wrong, that business is left out of the answer, regardless of its quality.
of travelers in Booking.com’s 33-country survey said they use AI to find local experiences or cultural activities.
Booking.com Global AI Sentiment Report, 2025
year-over-year growth in AI referrals to U.S. travel websites in July 2026.
Adobe Digital Insights, 2026
What Wayfinder does
Wayfinder is how CRAI finds and fixes what AI assistants get wrong about a business. The working rule: verify, publish, re-measure.
Measure
What AI assistants tell customers about a business.
Identify
Which answers the business should and shouldn’t appear in.
Recommend
Fixes for the web evidence behind those answers.
Re-measure
Confirm each fix took hold.
How the baseline works
The baseline is a controlled test. Questions are agreed with each business at kickoff, and every answer used as evidence is archived.
questions per business
AI models
repeated trials
data points per business
What the analysis shows
- ▪How often the business is named
- ▪Which competitors are named instead
- ▪Which third-party sources the models rely on (TripAdvisor, Google, Yelp, booking sites, local media)
- ▪Where the models state facts about the business incorrectly
Every answer is recorded, dated and open to the business to inspect.
What we’ve seen so far
From CRAI’s first two Wayfinder runs, with lodging properties in Southwest Colorado. These are why action lists target specific questions, not general visibility.
Visibility is per question, not per business
A property can appear in every “best B&B” answer and in none of the answers for the weddings or groups it serves.
Competitors fill the gaps
Where a business is missing, the models name competitors instead, often with mistaken facts.
Models respond to what a business publishes
Specific facts such as capacity, hours and amenities move answers within days.
The fixes are ordinary web work
Structured data, FAQ content, consistent listings and corrected third-party facts: work local consultants can carry out.
How an engagement runs
CRAI coordinates and verifies. Local consultants implement the fixes, and the businesses take part through an interview, access to their site and listings, and review of the findings.
Kickoff and fact verification
A 60-minute interview with each business to confirm the facts, name market opportunities and threats, and agree on the question set.
Baseline measurement
CRAI runs the controlled test across four AI models for every participating business, plus a community-wide view of the sources the models select.
Findings and action lists
Each business and its consultant gets the findings and a prioritized action list, with the evidence behind each item.
Implementation and closing report
Consultants carry out the highest-priority items. CRAI verifies each change and delivers the closing report.
Then, re-measurement. Some fixes show up in AI answers within days. Others may need several weeks, so a re-measurement once they have had time to propagate is what shows what moved.
What it delivers
Wayfinder starts with the businesses a visitor is most likely to ask an AI about, because what the AI says about them shapes how it describes the town. Each business gets a re-measurement showing how its changes affected its visibility. The town gets its first measured picture of how AI describes it.