Design Your AI Around the Guest — Not Around a Labor Problem

Last month a hotelier asked me a question I keep hearing in one form or another: “If I bring in AI, how many people do I let go?”

Wrong question.

I get why it comes up. Open any trade publication this week and the framing is everywhere. Skift ran a piece asking whether AI will fix travel’s labor problem. Hospitality Technology told operators to stop chasing automation targets and design AI around the guest experience instead. EHL published a clean way to think about it: decide what to automate, and what to keep human. Three respected voices, same week, circling the same idea.

They’re circling it because the labor math is brutal. In AHLA’s Front Desk Feedback survey, 76% of hotels reported staffing shortages — and even after raising wages, 79% still couldn’t fill their open roles, averaging seven vacancies per property. You are not going to out-hire this. Nobody is.

So the instinct is to reach for AI as a headcount lever. Cut here, automate there, hit a number.

That’s exactly the trap.

AI is not a way to have fewer people. It’s a way to do the work people were never able to do in the first place.

The gap hiring was never going to close

Think about what actually loses you a booking. A family messages your Instagram at 11pm — three rooms, four nights. Your front desk is closed. By the time someone sees it Monday, they’ve booked on an OTA and you’ve paid 18% commission on a guest who wrote to you first. That’s not a staffing failure. No sane owner staffs the front desk at 11pm on the off chance a DM comes in. It’s a coverage problem — a gap no amount of hiring was ever going to close.

And the cost of that gap is measurable. The classic MIT Lead Response Management study found you’re 21 times more likely to qualify a lead if you respond within five minutes instead of thirty. Harvard Business Review put an 80% drop in lead quality after that first five-minute window. And 78% of customers buy from whoever responds first. A guest never writes to just one hotel — they write to three or four. The first one back has the highest odds of the booking, every time. Speed isn’t a nice-to-have. It’s the single biggest determinant of who wins the guest.

No human team responds in seconds, in seventy languages, at 3am, and then follows up three days later with the guest who went quiet. Not because they’re not good — because they’re human, and they sleep, and they’re already stretched across seven open roles.

What to automate, what to keep human

Automate the work humans can’t reliably do: the instant first reply, the qualification, the professionally formatted proposal with the right upsells, and the patient follow-up with every indecisive guest — the indecisos, the ones who almost booked. This is repetitive, round-the-clock, never-forget work. It is exactly what machines are good at and people are not.

Keep human the work only humans can do: the welcome at the door. The judgment call when a guest arrives upset. The warmth that is the actual product of hospitality. AI should never touch that — and if you design it right, it never has to, because it hands the guest over already qualified, already delighted that someone answered at midnight.

That’s the whole point. You don’t design AI around a labor target. You design it around the guest — so the machine absorbs the conversation work, and your people are freed to do the human work they were hired for and are great at.

And to be clear about the other side of the table: this isn’t a war on OTAs. They bring guests to the conversation; that’s real value. The opportunity is simply to convert the direct inquiries you already have before they drift to a channel that costs you 15–25%. Optimize both. The enemy was never the OTA — it was the silence.

So if you’re asking how many people AI lets you cut, you’re optimizing the wrong variable. Ask instead: which conversations am I losing right now, at night and on weekends, that no human was ever going to catch? Start there. Automate that. Keep the welcome human.