Part of AI for Hotels — practical AI for independent hoteliers.

A hotel inbox never really stops.

Before guests arrive, they want to know what time they can check in, whether they can arrive early, where to park, if breakfast is included, whether the dog can come, and whether they can add another night.

While they are staying, it is the Wi-Fi code, what time breakfast finishes, whether they can have a late checkout, where to eat tonight, and could someone please look at the radiator in room 12.

After they leave, it is invoices, receipts, the phone charger left in the bedside drawer, and when the deposit comes back.

Almost every one of those emails is simple. That is the point. None of them is difficult, and that is exactly why nobody ever counts them.

But somebody still has to read it, work out what is actually being asked, check the booking, find the right answer, write the reply, send it, and sometimes chase it. A few minutes each. Multiply that across thirty occupied rooms, seven days a week, fifty-two weeks a year, and you are running a significant administrative operation that has never appeared as a line on your P&L.

What That Inbox Actually Costs

Take a realistic example: a 40-bedroom independent hotel running at around 75% occupancy, so roughly 30 occupied rooms a night.

A hotel like that might handle in the region of 90 guest communications a day across email and messaging — pre-arrival questions, booking queries, parking, check-in, breakfast, invoices, cancellation questions, room requests, late checkout, local recommendations and post-stay follow-ups.

That figure is illustrative. Your hotel might be at 50, it might be at 150, depending on your guest profile, how much of your business comes through OTAs, and how good your pre-arrival information already is. The shape of the calculation matters more than the number.

If a typical guest email absorbs around three minutes of staff time end to end:

90 emails × 3 minutes	270 minutes
Per day	4.5 hours
Per week	31.5 hours
Per year (52 weeks)	1,638 hours
At £15 per hour, all-in	£472.50 per week
Annualised	£24,570

That is close to a full-time role, spread invisibly across your reception desk, your duty managers and quite often yourself — consumed entirely by answering the same questions over and over.

I want to be clear about what this figure is and is not. It is not money you can remove from the business by installing software. Your reception desk still has to exist. It is a measure of how much repetitive administrative work is buried inside your operation, and how much of your team’s attention is going to tasks that do not need judgement.

What AI Actually Changes

A properly implemented AI communications assistant can read an incoming email, work out what the guest is asking, identify the property and the booking, pull the approved answer from your own information, write a reply in your hotel’s voice, and send it — in seconds, at three in the morning, on a bank holiday, without getting tired or inconsistent by the end of a long shift.

Here is what that looks like in practice.

Traditionally: it lands in the inbox. At some point somebody reads it, checks the arrival arrangements, works out what to say about the late arrival, remembers the parking situation, writes the reply, sends it. Maybe within the hour. Maybe not until the morning shift, by which point the guest is already on the road and has phoned instead.

With AI: it recognises the arrival-time question and the parking question, identifies the booking, retrieves the approved answers to both, and replies within seconds. No member of staff is involved at all — unless there is something unusual about that booking, in which case it gets escalated.

Automate the Repetition. Escalate the Exception.

This is the part I care most about, because it is where most AI projects in hospitality go wrong. The goal is not to remove people from guest communication. It is to stop people spending their day on the traffic that never needed them.

I work to three levels.

Level 1 — AI Handles It

Check-in and checkout times. Parking. Wi-Fi. Breakfast times and what is included. Directions. Standard policies. Basic room questions. How to get an invoice. The questions you answer fifty times a week.

Level 2 — AI Handles It Within Defined Rules

Early check-in. Late checkout. Extra nights. Room upgrades. Payment links. Booking amendments. Logging a maintenance issue. These need availability, pricing or a system to be checked — so AI operates inside boundaries you set, rather than deciding for itself.

Level 3 — A Human, Always

Serious complaints. Refunds. Compensation. Anything touching safety, security or a legal threat. Major maintenance failures. And anything at all where the AI is not confident.

Get those three levels right and you have an inbox that runs itself for the routine traffic and hands your team the things that genuinely need a person.

The Part Most Owners Have Not Considered

Reducing administration is the obvious benefit. It is not the biggest one.

Every day, guests email your hotel with questions that are, if you look at them properly, buying signals — and your team answers them as though they were administration.

A guest asks: “Can we arrive around midday?”

The standard reply: “Check-in is from 3pm.”

Conversation over. Nothing sold. The guest, who has just told you they want your product four hours earlier than they are entitled to it, goes and sits in a coffee shop.

Same question. Same effort. Completely different commercial outcome. And this is the kind of consistency that is genuinely hard to achieve with people: it is not that your receptionist does not know how to upsell, it is that at 4pm on a busy Friday with a queue at the desk, nobody upsells an email.

An AI assistant does it every time, at the same standard, on every shift.

Where the Opportunities Sit

• Breakfast. If it is not included in the rate, every breakfast question is a sale you are not asking for.
• Dinner. If you have a restaurant, “can you recommend somewhere for dinner tonight?” should never be answered with a list of other people’s restaurants.
• Early check-in. Any question about arrival time is a potential paid early arrival.
• Late checkout. Any question about checkout time is the same opportunity in reverse.
• Room upgrades. Pre-arrival contact is the cheapest upgrade channel you have — the guest is already committed and already thinking about the stay.
• Extra nights. If somebody asks about extending, availability can be checked and offered while they are still interested.

An Illustrative Revenue Model

Using the same 40-bedroom hotel, and assuming that of those 90 daily communications around 40 carry some kind of commercial angle:

Enquiry type	Per day	Converted	Average value	Daily
Arrival time / early check-in	8	15%	£20	£24.00
Checkout time / late checkout	6	15%	£25	£22.50
Breakfast	10	25%	£12	£30.00
Dinner / restaurant	9	20%	£27.50	£49.50
Room / upgrade questions	5	12%	£35	£21.00
Extending the stay	2	10%	£115	£23.00
Total	40			£170.00

£170 a day. £1,190 a week. Approximately £61,900 a year.

Look at the conversion rates in that table. None of them is ambitious. One breakfast question in four. One arrival-time question in seven. The model is not built on aggressive selling — it is built on asking, consistently, on enquiries you are already receiving and already answering.

This is a worked example, not a forecast. It illustrates how small conversion improvements compound across a year, and your actual result depends on your occupancy, your guest profile, what facilities you have to sell, your pricing, your availability, what you already convert today, and how well the AI is integrated with your PMS. If you have no restaurant, one line of that table disappears. If your rate already includes breakfast, another one does.

The precise figure is not the point. The point is that AI in a hotel inbox is not only a cost-reduction exercise. It can sell.

Two Numbers, and Why You Must Not Add Them Together

So the same 40-bedroom hotel gives us:

• ~£24,570 a year of administrative workload, expressed as a labour cost
• ~£61,900 a year of potential additional sales

I have seen these two numbers added together and presented as an £86,000 improvement. They cannot be. One is an illustrative measure of labour absorbed by repetitive work — much of which you will still be paying for, because you still need a reception desk. The other is potential sales, not profit — before the cost of the breakfast, the food cost on the dinner, the housekeeping on the upgraded room and the commission on the extra night.

What is fair to say is this. The traditional question about AI in hotels is “how many staff hours can we save?” It is the wrong question, and it is why so many owners look at it, decide it is not worth the disruption, and move on.

The better question is: how much repetitive administration can I remove, and how much additional revenue can I capture, from exactly the same conversations?

Why This Matters More for an Independent Hotel

A large group can spread revenue management, reservations, customer service, marketing, technology and administration across hundreds of properties. Somebody’s entire job is answering pre-arrival emails, and somebody else’s entire job is making sure that upgrades get offered.

You do not have that. In an independent hotel the same person is checking guests in, answering the phone, replying to Booking.com, working through the inbox, taking payments, producing invoices, handling a complaint, giving directions to the castle, selling breakfast and logging a broken shower — often in the same fifteen minutes.

That is the real argument for AI in an independent hotel. It is not about being cutting-edge. It gives you a communications capability that used to require a much bigger team, at a cost a 40-bedroom hotel can actually justify.

24/7 Cover Without 24/7 Administration

Guests do not email during office hours. They email late at night when they are booking, early in the morning before they travel, at weekends, from airports, and twenty minutes before they arrive.

Routine questions can be answered immediately, at any hour, while your team is asleep or busy elsewhere. People stay involved where human judgement genuinely adds something. Response times improve, answers stay consistent, and the guest who wanted to know about parking at 11pm gets told about parking at 11pm.

Less Noise Reaching the Owner

There is a quieter benefit that owners tend to appreciate more than they expect.

At the moment you are probably copied into a great many small operational conversations, on the reasonable basis that you would rather see things than not. The result is that the genuinely important messages are buried among dozens of trivial ones.

A well-configured system changes the shape of what reaches you:

That is a far more useful thing for an owner to receive than 52 messages.

How to Pilot This Without Starting a Technology Project

This does not need to be a systems overhaul, and I would advise against treating it as one.

1. Start with one property.
2. Analyse seven days of real guest communications. Not what you think you receive — what you actually receive.
3. Identify the repetitive questions. In most hotels a small number of question types account for the majority of the volume.
4. Build an approved knowledge base. Your policies, your answers, your tone, signed off by you.
5. Define the boundaries. What AI can answer, what it can offer within rules, and what always goes to a person.
6. Connect one channel first. Email is usually the sensible starting point.
7. Run it with human oversight. Everything visible, everything reviewable, nothing sent that you have not agreed the shape of.
8. Measure it. Response time, automation rate, escalations, errors, hours released, upsell conversion, revenue generated.

Then decide whether to widen it. If the numbers are not there, you have spent a few weeks finding that out on one property rather than committing to a platform.

If you want a starting point you can work through yourself, the Independent Hotel Owner’s AI Playbook covers the same ground in a practical, step-by-step format.

Where I Fit In

I do not sell software.

I advise independent hotel owners on operations, profitability, staffing and systems, and my interest in AI is entirely practical: it is one more tool for getting a hotel to run better and make more money. My job is to work out where it actually makes commercial sense — the same way I approach every other AI opportunity in a hotel.

Sometimes the answer is automation. Sometimes the answer is that the process itself is wrong and automating it would just make a bad process faster. And sometimes the honest answer is that AI is not the right tool for the problem you have got — in which case I will tell you that, because recommending it anyway helps neither of us.

You can see how I approach this kind of work in my case studies.