Dispatches from the front desk, the back office, and the phones that ring at 2:14pm.
The party-size cap on your booking form is not a mistake. It is a rational revenue decision implemented with the wrong tool. The party-size cliff is the point where the highest-value booking on the property turns into an email nobody answers.
In-room dining runs as a separate business, with its own card, its own extension, and often its own payroll, ordering from a restaurant the hotel already runs and already measures. Call it the phantom venue: 1.2% of revenue, a line no benchmark report breaks out, and Hilton quietly removed it in 2013 without ever naming it.
Every system a venue owns treats an event inquiry as a message: captured, routed, queued. It is actually a file, a stateful object with dates, headcount, budget, and a proposal. The live file is what happens when the inquiry arrives already advanced.
Per-minute billing quietly pays your voice AI vendor to cut the call short. Cost per Captured Call, the companion metric to Call Capture Rate, reprices the phone in the only unit that matters: bookings.
Every hotelier lived through the OTAs taxing the guest relationship. Now agentic AI is shaping up as the new toll booth. The honest answer: the threat is real, and the exit has exactly two parts.
Every hotel AI will eventually give a wrong answer. The metric that matters is not whether your vendor has a knowledge editor. It is Time to Correction: how long the wrong answer stays live.
Voice AI vendors advertise 97% automation without ever defining the denominator. We propose the metric they won't publish: the Honest Automation Rate.
Guest interfaces have climbed a ladder for thirty years: website, chatbot, voice, the live call. Every rung fixed a flaw and introduced a new one. The top rung, what the guest hears and sees moving in lockstep, is still empty.
Every hotel already runs a living knowledge base. It is called the front desk, and nothing else in the distribution stack can read it. A field note on knowledge half-life, the metric hospitality has never measured, and why it becomes existential the moment an AI agent starts speaking for your property.
Every hotel voice AI demo now sounds the same. Four questions separate the vendors that built a system from the vendors that built a demo: who holds the controls, what the guest sees, what the meter charges for, and whether you can watch it fail.
The next year of voice AI won't be won by the best model. It'll be won by what the model is wearing. A field note on the new category nobody is naming yet: the voice agent harness.
Frontier LLMs cap at 60-75% on multi-step tool use. The product question isn't whether models are smart. It's whether the stack around them is reliable.
Contacting a web lead inside five minutes makes it 21× more likely to qualify. At hotels, 39% of callers never reach a human. Math follows.
The hotel stack captures 50 data fields per stay and acts on three. Personalization isn't a data problem. It's a signal-selection problem.
Every major PMS deploys in weeks to months. FlowStay ships in a day — and the deployment gap is a structural wedge, not a feature.
Every canonical hospitality philosophy, read carefully, is a workaround for the same bottleneck: staff memory doesn't scale. FlowStay is that memory layer.
At full-service hotels, rooms are half the wallet. F&B, spa, and app-driven service are the other half, and the software still optimizes the first mile.
E-commerce industrialized cart recovery a decade ago. Hotels, facing 81% abandonment, still send one email and hope. The recoverable TAM is $40–92B.
Hotel voice AI quietly optimized for the only call type the industry measures: reservations. That's one in five. FlowStay was built for the other four.
AI won't replace your front desk. It will finally let them do the job they were hired for. The FlowStay operating principle.
Up to 62% of calls at independent hotels go unanswered. Your phone line is the highest-intent funnel you own, and it's being silently privatized.
Every independent hotel already runs an AI product. It's made of sticky notes, Gmail stars, and one overworked GM. Field notes from the back office.
OTAs captured 61% of independent-hotel bookings in 2024. The exit isn't a better banner ad. It's a new metric: Net Distribution Cost.
Independent hotels don't have a loyalty problem. They have a memory problem. The average returning guest has 2.3 profiles inside the PMS alone.
No posts in this category yet.