Guest messaging

Guest messaging platforms look much the same in a demo. They differ in what happens after the message arrives: how much gets answered without anyone typing, who picks up the rest and how fast, whether the system can read and write your booking data, and what it lets you measure afterwards. Those four things decide the result, and none of them is a feature you can tick off a list.

That is why this runs well only when a property treats it like any other part of operations: someone owns it, the rules for what gets automated are written down, and it is measured against money rather than speed. The technology is the easy half.

This is the high-level version of what we have written on the subject, drawn from running it across more than 190 properties. Each section links to the article that goes deeper.

Guests message on WhatsApp, so start there

Across the 190+ properties vGuest works with, 97% of guest conversations arrive on WhatsApp. Instagram, Messenger, Telegram and web chat matter, but they are the long tail. A messaging plan that treats WhatsApp as one channel among five is planning for the demo rather than for the guests.

WhatsApp guest messaging for hotels covers what works on the channel, what breaks, and how to measure it. Two practical details decide what a hotel can do there. The verified badge, covered step by step in How to get the WhatsApp blue badge for your hotel number, gates more than trust: it also switches on features such as WhatsApp group chats for hotel guests, which fit a family or a small travelling party, not an event.

Guests also do not write the way a form expects. They send voice notes, and they write in languages nobody on the night shift speaks. Voice Notes and Languages You Do Not Speak is about handling both without a scramble at 9pm.

The shape of the channel changes with the property. Beyond Check-In covers how it differs between a resort, a city hotel and a boutique, and what a hotel guest messaging platform should actually do is set out stage by stage, across the whole stay.

Most messages are the same few questions

Ask a team what it does all day and the answer is about service. Look at what it actually types and the list is short and repetitive: parking, check-in times, opening hours, availability, the same questions dozens of times a day.

That repetition is the case for automation and also the map for it. Audit the Questions Your Team Answers Forty Times a Day is the exercise to run before anything else. How one attraction cut 70% of phone inquiries shows what happens at an attraction when that list moves from the phone queue to messaging, and why the water park phone never stops turns it into a method for water parks and thermal baths, where one question can arrive three hundred times on a Saturday.

Across the portfolio, around 70% of the replies guests receive are sent by the assistant. That number is only as good as what the assistant knows, and very little of what it needs is technical. What You Actually Need Before an AI Assistant Goes Live lists it. Brief Your AI Assistant in Plain Language, Not Settings covers the part where most set-ups stall.

An assistant is accurate on the day it launches, and that is the last day anyone can assume it. Who Owns the Knowledge Base When It Goes Stale? is about the owner nobody appoints. The AI-Ready Hotel Org Chart covers the roles that form around that ownership as the channel grows.

The decision that matters is what the assistant must not answer

The most important configuration choice is not what the assistant answers. It is what it must never answer on its own: group bookings, accessibility, complaints with money attached, anything where the words understate what is being asked. The incident that ends a deployment is almost always on that list.

Group Bookings, Accessibility and the Requests You Must Not Automate is the list to write before switching anything on. From Guest Message to Ticket covers what happens next: routing the request to the right person, with the conversation attached, so the guest does not start again. What an AI concierge can book and what it cannot draws the same line from the other side, starting with the requests that should never be answered automatically.

Where the line sits is still argued, and the articles disagree on purpose. Human-in-the-Loop vs. Fully Automated Guest Messaging argues for the blend. When Automation Goes Too Far argues for firm limits. The Most Human Hotel Assistant Isn't Human at All argues that an assistant that is always there is often kinder than a stretched night shift.

Complaints deserve their own handling, because they are the highest-signal messages a property receives and the easiest to bury. The 2025 Guest Whisperer is about turning them into recovery rather than a review.

Messaging is only as good as the systems it can read

A reply about availability is only as reliable as the source behind it. A manually typed list drifts from the day it is written, fastest when accuracy matters most. The answer is to read the system of record, not a copy of it.

Can the Bot Just Check the Calendar? covers this for attractions, which ask a different first question than hotels do. For hotels the system of record is the PMS. vGuest connects to Optima, EzGo and MiniHotel, and to Mews, Cloudbeds, Guesty and SimpleBooking, and the integration has to write back as well as read.

Integration also fails in small places nobody tests. The Payment Field That Quietly Kills Your Digital Check-In is one example: a single form field that stops guests at the last step. When the fix sits with the vendor rather than with you, how to write a vendor ticket that actually gets fixed is what separates a resolution from a queue position.

Operators with more than one site hit a structural question early. One Assistant or Many? is worth answering deliberately rather than accepting a default. Smaller properties choosing tools from scratch should start with How to Pick the Right Tech Stack for a Small Hotel or Boutique Property. At that size the integration question has a different answer: Guest messaging for a 50-room hotel covers what changes below 100 rooms, including when a PMS connection is not worth the setup.

Measure outcomes, not response time

Every messaging dashboard leads with response time. It is easy to measure, it always improves when an assistant goes live, and it says almost nothing about whether the deployment was worth it.

Measuring Guest Communication Against GOPPAR, Not Response Time makes the case for the owner's number. The Metrics That Actually Matter in Guest Communication covers which figures flatter and which inform. The Hidden Cost of Friction in Guest Communication is the argument for counting the small delays, because guests do. From OTA Dependence to Direct Bookings covers where messaging can move revenue rather than cost.

Before any of that, a trial has to prove something. A pilot is not a trial of the software; it is a trial of whether the property can run it. What a 90-Day Guest Messaging Pilot Should Prove sets out what to decide before the first guest message.

What it costs

Three costs sit under guest messaging, and none of them is the number on a pricing page.

The first is the platform. Tier names like Starter and Professional rarely describe what drives the price, which is usually volume, channels and integration work. Hotel Chatbot Pricing explains what actually sets it.

The second is the channel itself. Meta bills WhatsApp business messages individually, and from 1 October 2026 service replies and utility messages inside the 24-hour window stop being free, according to Meta's pricing documentation (read 13 September 2026). Budget per message, and check the current rate card for your guests' countries; the WhatsApp pricing calculator for hotels does that arithmetic by guest country.

The third cost appears later and is the one people forget: leaving. Switching Guest Messaging Providers Without Losing Your History starts with the question that decides whether switching is possible at all.

How to choose a platform

The phrase "hotel AI assistant" covers two products that share a demo and almost nothing else. Pre-Booking vs Full-Journey Assistants explains the split, and buying the wrong one is the most expensive mistake in the category. How to Simplify Guest Communication Across Booking Channels compares the two kinds of platform honestly, including where competitors are the better choice.

The technology underneath matters too. Scripted Bot or LLM Assistant shows how to tell in three questions whether you are buying what failed in 2019 or what works now. From Scripted Bots to Smart Assistants is the history of why the old menu bots damaged trust in the category. Not all "Hotel AI" is created equal separates the kinds of model sold under one word, and what "AI-powered" actually means on a hotel tech invoice asks the same question of the bill.

For the evaluation itself: WhatsApp AI for Hotels: What to Compare Before You Buy covers the differences that only show in the second month. Twelve Questions to Put to Any Guest Messaging Vendor gets past the demo. The Hotel Guest Messaging Glossary decodes the vocabulary vendors use, including where a term hides a question you should ask.

Some of the messages will come from AI assistants

A growing share of requests will not be typed by the guest. They will be sent by the guest's own assistant, precise and several questions at once: early check-in, parking, the pet policy, a spa slot. Your Guest's AI Will Message You Before They Do sets out the shift, and Answering an AI Agent covers what the messaging has to handle when it happens. Some of those trips are now booked inside the assistant itself: Google AI Mode now books hotels covers who the guest contacts afterwards, which is still the property.

Those assistants read the property before they write. Your Rooms Rank. Your Policies Do Not. is about the answerable questions, such as dogs, cancellation and parking, that most hotel sites leave unanswered. The AI Protocol Shift argues for treating machine-readable guest information as operations rather than marketing.

Where to start

Spend one week counting the questions your team answers by message and by phone, and write down the requests the assistant must never answer on its own. Those two lists decide most of what follows: what to automate, what to route to a person, and what to measure. They are also the best brief any vendor will receive from you.