This insight summarises Beyond Tokenomics: The Threat of the Invisible Hotel by Patrick Burkhardt, Senior Director of Sales at Curacity, published on Hospitality Net.
Reframe the question. Everyone is arguing about how expensive AI queries are. Burkhardt argues the more important question is: does the AI name you at all? Recommendation inclusion — not query economics — determines whether you have a market.
The data behind the alarm:
- Yext analysed 6.8M hospitality citations — Gemini, ChatGPT, and Perplexity each trust different source sets. No unified visibility strategy works across all three.
- The 5W AI Visibility Index shows citation share concentrating around the top three brands in most competitive sets.
- Loyalty program size did NOT predict citation levels. Being big isn't enough.
- ~65% of hotel queries trigger live web searches — answers assemble in real time from dynamic sources, not from static training data.
Who actually controls visibility. Editorial coverage, expert lists, and third-party sources upstream of booking engines — not brand.com — increasingly decide who gets named in AI answers. Most hotel teams don't yet manage these relationships as a distribution channel.
The historical parallel: this is the OTA distribution war, replayed. Brands that mastered SEO in the 2010s and OTA relationships in the 2000s must now master AI recommendation layers — before the field saturates.
What to do about it:
- Treat citation share as a board-level KPI — alongside RevPAR and direct-booking share.
- Invest upstream, not just downstream — earned media, structured data, and expert-list placement now shape what AI cites.
- Move fast on structural authority. Early, disciplined investment compounds; latecomers pay entrenchment costs.
- Watch for signals — quarterly citation-share indices and "Head of AI Visibility" hires at major chains will mark when the market institutionalizes.
Bottom line: the invisible hotel is not the one paying too much for AI infrastructure. It's the one AI never mentions. Fixing that is a marketing and distribution problem, not a token cost problem.