Send a two-way reminder sequence (a confirmation at booking, a message about 48 hours before asking customers to confirm or rearrange, and an on-the-day message), let AI read the replies and offer new slots when someone can't make it, keep a short-notice list to refill gaps, and take deposits for costly slots. Most firms can halve their no-show rate this way.
Most no-shows aren't forgetfulness. They're customers who knew they couldn't make it but found it awkward or slow to say so, and bookings nobody ever confirmed. The fix is to make rearranging easier than not turning up. For trades there's a second kind: the "no access" visit, where the customer is out, the key-holder never got the message, or the engineer can't get in. Reminders have to reach the person who opens the door.
What your no-shows cost before you change anything
Put a number on it first, because it decides how much effort and money the fix deserves. For an illustrative ten-person HVAC installer doing about 120 booked service and repair visits a month:
- No-access or no-show visits last quarter: about 7%, so roughly 8 a month.
- Cost of each: about 90 minutes of an engineer's time including travel, at roughly $55 an hour fully loaded, plus around $15 in van costs. Call it $97.
- Monthly cost: 8 × $97, about $780, before counting the customer who then waits another week and the slot that could have gone to someone else.
If you don't know your rate, start logging now: every booked appointment, and whether it happened, was rearranged in time, was a late cancellation, or was a no-show. Two weeks of honest logging tells you more than any benchmark.
The reminder sequence that makes rearranging easy
| When | Channel | What it says | Why |
|---|---|---|---|
| At booking | Email and text | Date, time window, what to expect, how to rearrange | Catches wrong details immediately |
| About 48 hours before | Text | Asks the customer to reply YES to confirm or CHANGE to rearrange | The one that prevents no-shows; enough notice to refill the slot |
| Morning of, or on the way | Text | Arrival window or "engineer on the way", with a name | Prevents the "popped out to the shops" no-access |
| After a no-show | Text, then a call | A neutral "sorry we missed you" with two new slots | Recovers the job rather than losing the customer |
Wording matters as much as timing. The 48-hour message should make both answers easy. An example:
Hi [first name], your boiler service is on Thursday 14th between 8 and 10am. Reply YES to confirm, or CHANGE and we'll offer other times. Please reply either way so we can plan the engineer's route. [Firm name]
That's under 180 characters without the name fields, which matters: Calendly's text workflows limit messages to 180 characters, and Jobber's texts to 160. The phrase "so we can plan the engineer's route" gives a reason to reply, which tends to lift replies compared with a bare reminder. The no-show follow-up should be free of blame:
Hi [first name], sorry we missed you today. Would Tuesday 8-10am or Wednesday 12-2pm suit instead? Just reply with the day. [Firm name]
Where AI earns its keep: reading the replies
Reminders themselves don't need AI; any booking or job tool can send them. AI helps with what comes back. Customers don't reply "CHANGE". They reply "can't do thurs, fri any good? after 12 ideally", "yes but can he ring when he's 10 mins away", or "who is this?". A person reading 40 of those a week takes time; a simple AI step can classify each one and draft the response.
A classification prompt for an automation step:
You read customers' replies to appointment reminders for a heating firm.
Return JSON only:
intent: "confirm" | "confirm with request" | "reschedule" | "cancel" |
"question" | "unclear"
preferred_days: list of days or dates mentioned, or []
preferred_times: "morning" | "afternoon" | "any" | "not given"
request: any special request in 10 words or fewer, or ""
needs_person: true if the customer sounds unhappy, mentions a fault getting
worse, a vulnerable person, or anything you can't classify
Rules: "yes but..." is "confirm with request" only if the appointment still
stands. Never treat a question as a confirmation.An illustrative reply and the output:
Reply: "hi can't do thurs now, fri any good? after 12 ideally. also the
pressure keeps dropping, is that ok?"
{"intent": "reschedule", "preferred_days": ["Friday"],
"preferred_times": "afternoon", "request": "",
"needs_person": true}The output is right to flag a person: a pressure problem getting worse may change what the visit needs, and the office should ring rather than just rebook. What you'd check is that the automation doesn't then book Friday afternoon without the office seeing it. The rule this firm used: if needs_person is true, the reply goes to the office queue with the AI's suggested slots attached, and a person confirms. Otherwise, the AI offers the two nearest matching slots and books whichever the customer picks.
Automatic rebooking, step by step
Rebooking is where the no-show rate really falls, because a rearranged appointment is a kept appointment. The flow, which you can build in most job-management tools or with an automation platform connected to your calendar:
- Trigger: a "reschedule" reply, a "CHANGE" reply, no reply 24 hours after the 48-hour reminder, or a visit marked as no access.
- Find slots: the same appointment type, at least 24 hours away, matching any preferred day and time, and on a day the engineer is already working in that area where your tool knows it.
- Offer two, not five. Two options get a quicker decision than a long list.
- Hold briefly: keep the offered slots reserved for a couple of hours, then release them.
- Confirm and clean up: book the chosen slot, cancel the old one, and make sure old reminders are cancelled too. Jobber does this automatically when a visit is rescheduled; in a home-built setup it's the step most often forgotten.
- Refill the freed slot: offer it to the short-notice list.
For non-replies, don't cancel automatically. Send one more nudge ("We haven't heard back; we'll keep your slot unless you tell us otherwise"), and have the office ring anyone who still hasn't replied the day before. For a costly visit, a 60-second call is worth more than another text.
Which bookings deserve a phone call instead of a text
Not every appointment carries the same risk. Your own history shows which ones go wrong, and an AI assistant can find the pattern if you give it a few months of bookings with their outcomes. Export booking date, appointment date, customer type (new or returning), how the booking was made, whether a reminder was confirmed, and whether the visit happened. Then ask which combinations had the highest no-show rates.
For the heating firm, three flags stood out: first-time customers who booked more than three weeks ahead, anyone who had missed a visit before, and bookings made by someone other than the occupant, such as a landlord for a tenant. Visits with any of those flags got a personal call from the office the day before, on top of the texts. That was about one visit in six, a manageable 15 minutes of calls a day, aimed where it mattered. Keep the rules simple and visible; a list of three flags the office understands beats an opaque score nobody trusts.
A short-notice list that refills the gaps
A freed slot two days out is worth something only if someone can take it. Keep a list of customers happy to be seen sooner: people waiting for non-urgent work, customers who asked for "earliest available", or those who've said they're flexible. When a slot frees up, the automation texts the list in order, and the first to reply gets it:
Hi [first name], we've had a slot come up on Wednesday 8-10am for your service. Reply YES to take it and we'll move your booking forward. First to reply gets it. [Firm name]
Two rules keep it fair: only offer slots of the right type and length, and cap the list message to four or five people at a time so you don't disappoint twenty. The same idea underpins class waitlists and salon gap-filling; how salons refill diary gaps with AI shows it in a different setting.
Deposits and fees: where they help and where they hurt
Charging for missed appointments works, but it changes how customers feel about booking, so use it where the slot is expensive and the customer has chosen it freely.
- A kitchen fitter's design consultation ties up a designer for 90 minutes. A modest fee at booking, refunded against any order, filters casual bookings and makes people turn up. Google Workspace Business Standard's appointment schedules can take payment through Stripe at booking, and Calendly has payment integrations.
- A surveying firm commonly takes the full fee at booking, which makes no-shows rare but shifts the problem to access (below).
- An HVAC installer's service-plan customers have already paid for the visit, so a no-show fee risks the renewal. Better reminders and easier rearranging are the right tools there; a fee might suit one-off call-outs instead.
Whatever you charge, put the terms in the booking confirmation in plain words, and check consumer rules where you operate before introducing fees; an adviser can tell you what's enforceable. The salon tutorial on cutting no-shows with reminders and deposits goes further into deposit wording.
Tools, and what each can and can't send
Check what your existing software does before buying anything, because most booking and job tools include reminders. Some details that trip people up:
- Jobber sends visit reminders by email and text on selected plans, timed relative to the visit or at a set time each day. Customers get a link to the client hub, where they can press "Confirm Appointment". Rescheduling automatically cancels reminders for the old time. Text reminders are only available in some countries, so check yours.
- Calendly sends text reminders through Workflows on its paid plans, but not during the free trial. Invitees must opt in to texts and you can't make the phone field compulsory; new accounts can only use built-in text templates for a while, to prevent spam.
- Microsoft Bookings sends email confirmations and reminders on ordinary Microsoft 365 business plans. Its text notifications need a Teams Premium licence and only reach numbers in a few countries, so many small firms send texts another way.
- Google Calendar appointment schedules send email reminders, but not texts.
- A business phone system plus automation, such as a VoIP number with an automation platform, can send texts from any booking tool and pass replies to an AI step. More setup, but it works with whatever calendar you use, and replies arrive in the same place as the rest of your customer texts.
If you're choosing a booking system at the same time, AI appointment scheduling tools and setup covers the booking side before the reminders.
The heating firm's first two months
The HVAC installer already used job-management software with visit reminders switched on, but only as a one-way email the day before. The changes they made:
- Switched reminders to text at 48 hours with the YES or CHANGE wording, keeping email as a backup.
- Added an on-the-way text sent by the engineer's app, with a 30-minute warning.
- Connected replies to an AI classification step that drafted responses and offered two slots for reschedules, with the office approving anything flagged.
- Started a short-notice list of 25 customers waiting for non-urgent work.
- Had the office ring any customer who hadn't replied by the afternoon before.
Illustrative results after two months: no-access visits fell from about 7% to about 3%, roughly five fewer wasted visits a month, worth around $480 at the earlier cost figure. About two-thirds of rescheduled slots were refilled from the short-notice list, often with customers who were delighted to be seen sooner. The office spent about 20 minutes a day on flagged replies and non-reply calls, less than it had spent ringing customers to confirm before. The automation and AI step cost a small monthly amount on the firm's existing automation plan.
When the person booking isn't the person at the door
Surveyors, and trades working on rented or empty properties, hit a no-access problem reminders to the customer don't fix. A six-person surveying firm found that most of its failed inspections weren't the buyer's fault at all: the seller wasn't told, the estate agent's keys weren't available, or a tenant didn't know anyone was coming.
The fix was to capture an "access contact" as a separate field at booking (the agent, the seller or the tenant) and send the 48-hour and on-the-day messages to that person, with the buyer copied on the confirmation. The access reminder reads differently:
Hello, this is [firm name]. A surveyor is booked to inspect [property] on Tuesday 12th at 10am on behalf of the buyer. Please reply YES to confirm access, or tell us if the time won't work.
An AI step reading access replies needs one extra instruction: a reply like "keys at the office from 9" is a confirmation with a request, and should add a note to the surveyor's job, not trigger a reschedule. After three months, the firm's failed inspections fell by more than half, mainly because access problems surfaced two days early instead of on the doorstep.
Mistakes that undo the gains
- Too many messages. Four reminders for one visit trains people to ignore them, and some opt out entirely. Two plus a confirmation is plenty for most visits.
- Stale reminders after a reschedule. A customer moves to Friday and still gets Thursday's reminder, then doesn't know which to believe. Check that rescheduling cancels old messages in your setup.
- Reading "yes, but" as yes. "Yes but I might be late back from work" is a warning, not a confirmation. The classification rule above helps, and so does a person reviewing flagged replies.
- Rebooking into impossible slots. An AI offering a slot across town an hour after another visit. Limit offers to slots your scheduling tool considers feasible, or have the office approve visit reschedules.
- Wrong numbers. A typo in the phone number means every reminder goes to a stranger. The confirmation text at booking is your check: if nobody ever replies to it, ring the customer.
- Blaming tone. "You failed to attend your appointment" loses customers. Keep follow-ups neutral and offer new times straight away.
Measure the same things each month: no-shows and no-access visits as a share of booked visits, late cancellations, the share of freed slots refilled, and office time spent on replies. If no-shows fall and office time doesn't rise, keep going. If replies pile up unread, the AI step needs tightening or the office needs a fixed daily slot to clear them. For the heating firm, reminding customers their annual service is due is a separate job from this one; that's covered in AI service plan reminders for heating and cooling customers.
No-shows and reminders: follow-up questions
How many reminders should I send before an appointment?
Two is right for most service visits: one about two days before asking the customer to confirm or rearrange, and one on the morning or an hour before arrival. A booking confirmation at the time of booking comes on top. More than that tends to irritate people and raises opt-outs without cutting no-shows further. Long-lead appointments booked weeks ahead can justify an extra reminder a week before.
Do I need customers' permission to send text reminders?
Usually you need a lawful basis and, for texts, many tools and jurisdictions expect a clear opt-in. Calendly, for example, requires invitees to choose to receive texts and won't let you make the phone field compulsory. Ask at booking whether customers want reminders by text, keep reminders purely about the appointment, and check the rules where you operate with an adviser if you're unsure.
Should the AI rebook customers without asking me?
Let it offer and confirm new slots within rules you set, such as only slots at least 24 hours away, only the same appointment type, and never double-booking a visit that needs travel. Keep a person in the loop for anything outside those rules, for repeat no-shows, and for customers who seem unhappy, where a phone call does more than an automated offer.
What no-show rate should I aim for?
Compare against your own baseline rather than a published figure, because rates vary widely by trade, customer type and how far ahead bookings are made. A realistic goal is to halve your current rate within three months of adding two-way reminders and easy rearranging. If you can't measure your current rate, start logging no-shows and no-access visits now, before changing anything.
Further reads
- Can AI Run Your Studio's Class Waitlist Automatically? — Waitlists that refill cancelled places automatically.
- How Mobile Dog Groomers Use AI for Routes, Reminders and Rebooking — Reminders and rebooking that respect travel routes.
- Can AI Help a Cleaning Company Rebook More Regular Clients? — Rebooking regular clients in a service business.
- How Dental Practices Use AI for Recalls and Appointment Reminders — Recalls and reminders in a regulated setting.
- AI Job Scheduling and Dispatch for Trades and Field Service Teams — Rebooking that has to respect routes and crews.
- Can AI Cut Last-Minute Driving Lesson Cancellations? — Cutting last-minute cancellations for one-to-one lessons.
- How to Map Your Customer Journey and Find Where AI Helps — A journey grid, evidence sources and the signs a step suits AI, walked through with a new puppy owner's first year at a veterinary practice.
- AI Booking Systems for Appointment Businesses: What to Check — A 24-point checklist for choosing an AI booking system, grouped by risk, with how to verify each item and a two-hour test script to run before you sign.
- DIY or Get Help? Setting Up AI Bookings for a Small Business — Six questions that tell you whether to build AI bookings yourself, with three worked cases and a weekend plan if you go it alone.
- How Wedding Venues Use AI for Viewings, Enquiries and Follow-Ups — Use AI at three points in a venue's sales pipeline: the first reply, the notes after each viewing, and follow-ups timed around provisional holds.
- How Photographers Can Fill Mini-Session Slots With AI Marketing — A three-week, waitlist-first campaign for mini sessions: slot maths, past-client segments, copy prompts and honest ways to fill the last slots.
- How Tattoo Studios Handle Enquiries and Deposits With AI — How a tattoo studio can use AI to sort and answer enquiries, collect what artists need to quote, and take deposits safely through booking software.
- How Car Valeting and Detailing Firms Use AI to Fill the Diary — An empty-slot audit, rebooking intervals, photo-quote prompts and standby-list messages that help a valeting or detailing business keep its diary full.
- How to Use Booking Data and AI to Fix Your Class Timetable — Five numbers per class slot, a prompt that finds the patterns, and a six-week test so timetable changes don't cost you regulars.
- What Is an AI Receptionist and How Does It Handle Bookings? — A plain-English look at what happens inside an AI receptionist during a booking call, and the three very different things vendors mean by 'handles bookings'.
- 7 AI Mistakes Restaurant Owners Make With Bookings and Reviews — Seven ways restaurant AI goes wrong with tables and reviews, each with a real-looking example, what it costs and how to fix it.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Jobber help centre on assessment and visit reminders; Calendly help on sending text messages with Workflows; Microsoft documentation on SMS notifications in Bookings; Google Calendar help on appointment schedules.