If you run a small guesthouse in Japan, you know the feeling. The property itself is fine and the guests are mostly lovely. But your day is cut into small pieces: a message about the door code, a check of what nearby places are charging next weekend, a listing on one OTA that no longer matches the listing on another.

This is a product spotlight, so I’ll say it plainly: this post is about BenStay’s own short-stay apartment in Tokyo and the automation we built to run it. The first half is about the problem, though, and how anyone could approach it, whether or not you ever use anything we’ve built.

TL;DR

  • A small guesthouse operator’s time mostly goes to three repetitive jobs: answering guest messages, adjusting prices, and keeping listings consistent across several OTAs.
  • Most guest questions fall into a few stable categories: check-in, transit directions, luggage storage, house rules, and how to use the facilities.
  • Those answers are property-specific but rarely change, so a chatbot grounded in the property’s own information can answer documented, repeat questions instantly.
  • The questions that still need a human are usually reservation changes and refunds, so we built our system around that split.
  • At BenStay we run our own Tokyo short-stay apartment on guest chatbots, dynamic pricing, and multi-OTA listing management, and we’re upfront about where that fits and where it doesn’t.

Why Does Running a Small Guesthouse Feel Like Death by a Thousand Messages?

Running a small guesthouse is exhausting mainly because the work is constant, repetitive, and spread across too many places, not because any single task is hard. There are three main jobs:

  1. Guest communication. Questions arrive before arrival, during the stay, and sometimes after check-out, at any hour and in many languages.
  2. Pricing. Demand in Tokyo shifts with seasons, holidays, events, and the day of the week. A rate that was right last month can quietly lose you money this month.
  3. Listings across several OTAs. Many lodging businesses rely on OTA channels; a 2024 JATA survey found that 83% of lodging businesses surveyed used OTAs as an inbound reservation channel. If you list on more than one, every change has to be made more than once.

None of these is usually hard on its own; the difficulty is that they arrive all day, across several systems.

What Do Guests Actually Ask About?

Most guest messages fall into a handful of predictable categories. In our experience, the bulk of questions before and during a stay are about:

  • Getting in: the check-in procedure, door codes, and what to do if they arrive before check-in time
  • Getting there: directions from the nearest station or the airport
  • Luggage: where to leave bags before check-in or after check-out
  • House rules: trash separation, quiet hours, and shoes
  • Facilities: wifi, laundry, and air-conditioning

What matters about this list is that the answers are specific to the property but stable. Your door procedure and your trash rules don’t change from guest to guest. That makes them ideal for automation.

How Could Any Operator Approach This, Even Without Special Tools?

You can start by writing your answers down once and reusing them every time. Even with no software at all:

  • Build an answer bank. Go through your last few months of messages, group them by the categories above, and write one clear, complete answer for each.
  • Send answers before guests ask. Most check-in and transit questions disappear if the information reaches the guest at the right moment before arrival.
  • Set a pricing routine. Even a weekly review of upcoming dates, holidays, and local events is better than a rate you set once and forgot.
  • Keep one master copy of your listing. Treat one document as the source of truth for descriptions, rules, and photos, and copy from it to each OTA so they don’t drift apart.

This will get a solo operator surprisingly far. The limit is that it still depends on you being awake, available, and disciplined, every day.

What Did We Build for Our Own Property?

We built automation for exactly those three jobs: guest chatbots, dynamic pricing, and multi-OTA listing management. We didn’t start out wanting to build software. We run a short-stay apartment in Tokyo, and we got tired of being the bottleneck.

Guest chatbots. Our chatbot is grounded in the property’s own information, so it answers from what’s actually true about that building, not from generic travel advice. That covers the categories above: check-in, directions, luggage, house rules, and facilities. When the answer is stable and documented, the guest gets it instantly, whatever the time.

Dynamic pricing. Instead of one rate we update when we remember, prices adjust over time to reflect demand. That takes the “did I forget to raise rates for that holiday weekend?” worry off our plate.

Multi-OTA listing management. We list on more than one platform, so we manage the listings together rather than editing each one by hand. The aim is simple: what a guest sees on one OTA should match what they see on another.

Where Does This Fit, and Where Doesn’t It?

It fits best when the work is repetitive and the answers are stable. The chatbot is good at “how do I get in?” and “where’s the trash?” It is not a replacement for a person when the situation needs judgment.

In practice, the guest questions that still need a human are usually reservation changes and refunds. Those involve money, policy, and sometimes a guest who’s stressed. We don’t want a bot negotiating those, and we’ve designed our setup so those messages come to us.

Some other honest limits:

  • Automation is only as good as the information behind it. If your house rules are vague or out of date, a grounded chatbot will faithfully repeat vague, out-of-date answers. The work of writing things down clearly comes first.
  • Pricing automation still needs a human to look at it now and then. It handles the routine adjustments, but you still own the strategy.
  • This came out of one operation’s needs. We built it for how we run a Tokyo short-stay apartment. If your property or workflow is very different, some of it may not map neatly.

If you want to see how we run things, you can find BenStay at benstay.jp.

FAQ

Q: What kinds of guest questions can a guesthouse chatbot realistically handle?

Questions with stable, property-specific answers: the check-in procedure and door codes, transit directions, luggage storage, house rules such as trash separation and quiet hours, and how to use the wifi, laundry, and air-conditioning.

Q: Which guest messages should still go to a human?

Reservation changes and refunds. They involve money and policy decisions, and they usually need judgment and a personal touch.

Q: Do I need software to reduce my guest message workload?

No. Writing a clear answer bank and sending key information before arrival already cuts a lot of repeat questions. Automation mainly removes the need for you to be the one sending those answers around the clock.