Why five minutes, when Airbnb only asks for 24 hours?
Airbnb's rule is generous. Your response rate is the share of new inquiries and requests you answered within 24 hours over the past 30 days, and Superhost status needs 90% of them inside that window (Airbnb Help Center). Miss 24 hours and the guest can move on; Airbnb tells guests plainly that "hosts have 24 hours to respond" (Airbnb Help Center).
The five-minute target comes from the other side of the conversation. Airbnb's own hosting guidance says the most successful hosts reply as soon as a message arrives, that a good response rate can help a listing appear higher in search, and that speed matters most before booking and around check-in and checkout (Airbnb Resource Center). A guest comparing three listings at 9 pm books the one that answers at 9:04. A guest locked out at midnight remembers how long the door stayed shut.
So the question for a manager with 40 or 400 units is not "how do we stay inside 24 hours?" It is "how do we answer in minutes, at 2 am, in the guest's language, without a night shift?"
What stops a manager from answering fast?
Three things, in the order they usually show up.
Volume spread across channels. Airbnb is one inbox. Vrbo, Booking.com, WhatsApp, email, SMS, and the direct-booking site are six more. The operators we talk to describe it the same way: "too many channels, too many messages" and "switching between 6+ platforms daily." Reply time is not a staffing problem at first; it is a tab-switching problem.
Context that lives somewhere else. The answer to "what's the Wi-Fi password?" is in the PMS. The answer to "can we check in early?" depends on today's cleaning schedule. The answer to "is the pool heated?" is in a listing description someone wrote two years ago. The person holding the phone has to open three systems before they can type one sentence.
The night and the weekend. A team that answers in four minutes from 9 to 6 answers in four hours at 2 am. Guests do not check in on office hours.
Hiring a coordinator or a virtual assistant fixes the third problem and makes the first two more expensive, because every new hire needs the same three systems open. Cost scales with units. That is the trap this article is trying to get you out of.
Step 1: Put every channel in one inbox
Before any automation, collapse the channels. One thread per guest, with Airbnb, Vrbo, WhatsApp, email, and SMS in the same view, and the reservation and property attached to it. Read the Trellis AI Inbox page for how that looks in practice, and the Airbnb and WhatsApp integration pages for what each channel syncs.
The test of a unified inbox is simple: can the person answering see the check-in date, the door code, and last night's cleaning status without leaving the message? If not, the reply will take ten minutes regardless of how many people you hire.
Step 2: Let an agent draft every reply, and send the routine ones
This is where the five minutes come from. An agent that reads the reservation, the property details, and the conversation history can draft an answer to "what time is checkout?" in seconds. The decision is not whether it drafts, but whether it sends.
The pattern that works for the managers on Trellis is autopilot earned per channel. The agent starts in suggestion mode everywhere: it drafts, a human approves. On the channels where the drafts are consistently right, you switch that channel to autopilot and the agent sends on its own. Airbnb pre-stay questions usually earn it first. Owner communication and damage disputes usually never do.
Customers land at different points on that dial. One operator reviewed a weekend of suggested replies and would not have changed any of them, then moved the channel to autopilot. Another prefers to stay put: "We just prefer suggestions more than autopilot. Suggestions works good." Both are running the same setup.
How much of the volume the agent ends up handling varies by manager, channel, and property mix. The number that matters for your operation is the one on your own dashboard, per channel, and it should climb week by week.
Step 3: Write down what must escalate
Speed without control is how you end up with the agent telling a guest about a gym the property does not have. Every manager we onboard writes a short escalation list before flipping any channel to autopilot. A typical list:
- Any mention of a refund, a dispute, or a review threat.
- Safety issues: smoke, water, locks, medical.
- Anything about a specific property flagged as sensitive. One operator told us certain properties "should never ever be able to have AI responses," and the system respects that per property.
- Price changes or discounts the agent is not authorized to offer.
Everything on the list goes to a human with the full conversation attached. Everything else gets answered in the guest's language, in minutes, at any hour. The agent's job is to make the escalations rare and well-prepared, not to pretend they do not exist.
Step 4: Connect the reply to the work it creates
Half of the late-night messages are not questions; they are requests. "The heating isn't working." "There are no towels." "Can we get a late checkout?" A fast reply that says "we're on it" and then nothing is worse than a slow reply that gets the heating fixed.
The agent should be able to turn the message into a task for the right person: a maintenance ticket with the property, a note to the cleaner on shift, a hold on the calendar for the late checkout. That is the difference between messaging automation and operations automation, and it is why a guest-messaging tool alone does not get you out of hiring. See Trellis Operations for how a guest message becomes a task, and if your bottleneck is turnovers rather than messages, the Trellis vs. Breezeway comparison is the better starting point.
Step 5: Check accuracy every week, not every message
Once a channel is on autopilot, reviewing every message defeats the point. Instead, pick a sample weekly: 20 conversations per channel, read the agent's replies, mark the ones you would have written differently. Two questions for each miss:
- Was the information wrong (fix the property notes), or was the tone wrong (fix the agent's instructions)?
- Should this type of message have escalated (add it to the list)?
This is also how you decide when a channel is ready for autopilot in the first place. On Trellis, billing only starts when autopilot accuracy reaches 70%, so the weekly check is built into the commercial model rather than left to good intentions.
What this looks like at 50 units
A 50-unit manager without a night shift, after the five steps:
| Message type | Who answers | Typical reply time |
|---|---|---|
| Pre-booking questions (Airbnb, Vrbo) | Agent on autopilot | Under 2 minutes, any hour |
| Check-in logistics, Wi-Fi, parking | Agent on autopilot | Under 2 minutes |
| Maintenance reports | Agent replies and creates a task; human confirms dispatch | Reply in minutes, fix depends on the vendor |
| Refunds, disputes, sensitive properties | Human, with context attached | Next available person |
| Owner messages | Human, agent drafts | Office hours |
The headcount did not change. The reply time for the top two rows, which are most of the volume, went from hours to minutes.
Frequently asked questions
Does a fast reply from an AI count toward Airbnb's response rate?
Yes. Airbnb's response rate counts the new inquiries and requests your host account answered within 24 hours (Airbnb Help Center). The agent replies through your connected Airbnb account, so its replies are your account's replies for the purposes of the response rate and the Superhost requirement.
Can I keep a human approving every message on some channels?
Yes. Autopilot is a per-channel and per-property setting. Many managers keep owner messages and one or two sensitive properties in suggestion mode permanently while everything else runs on autopilot.
What happens when the agent does not know the answer?
It escalates to a person with the conversation and the reservation attached, and it tells the guest someone is on it. It should never invent an amenity or a price. When you find it did, the fix is in the property notes or the escalation list, and the weekly accuracy check is where you catch it.
Do I need to leave my PMS?
No. The agent reads the reservation and property data from the PMS you already run. Trellis is a Hostaway certified partner and connects to Guesty, Hostfully, Hospitable, OwnerRez, and the PMSs available through Calry.
Sources
All retrieved 2026-08-22.
- Airbnb Help Center, "Why hosts are asked to respond within 24 hours": https://www.airbnb.com/help/article/2414
- Airbnb Help Center, "Improve your response rate and response time": https://www.airbnb.com/help/article/430
- Airbnb Help Center, "What to do if a host isn't responding": https://www.airbnb.com/help/article/88
- Airbnb Resource Center, "Why it's so important to respond quickly to guests": https://www.airbnb.com/resources/hosting-homes/a/why-its-so-important-to-respond-quickly-to-guests-374
