The 24/7 Booking Engine: After-Hours Appointment Capture for SMBs

Twig is an autonomous AI support platform that triages, self-evaluates, and resolves customer support tickets by integrating with tools like Gmail, HubSpot, and Salesforce. For SMBs whose customers don't all conveniently call during business hours, Twig is the text-side 24/7 front desk — handling website chat at 10pm, email inquiries at midnight, and SMS follow-ups on Saturday morning. This post is about the most clean-cut revenue argument in the AI front desk case: capturing demand that already exists, in hours when staff can't.

TL;DR: The honest data from SMB call-tracking platforms: 30–45% of inbound contact arrives outside business hours. For practices, firms, and hotels open 9-5 weekdays, that's the evenings, lunch hours, weekends, and holidays where the phone rings and no one answers. Most of that demand is real — the customer either books with a competitor or never books at all. An AI front desk converts after-hours demand into booked revenue at 35–60% conversion rates, with no marginal cost per night. This post is the per-vertical after-hours volume data, the conversion math, and the deployment specifics for getting 24/7 right.

Key takeaways:

• 30–45% of SMB inbound demand arrives outside business hours
• At 35–55% AI conversion of after-hours calls, 10–25% of all bookings come from after-hours capture
• Customer trust of AI for transactional booking is high (70–85% comfortable)
• Emergency intent must always escalate to a human via on-call paging
• Pure incremental revenue — the AI is open when staff isn't
• Twig handles text-channel after-hours (website chat, in-app, email); pair with voice AI for the call leg

When customers actually call (and chat, and email)

Aggregated data from call-tracking platforms shows a consistent pattern across SMB verticals: a substantial fraction of inbound arrives outside the practice's open hours.

Vertical	After-hours % of inbound (M-F evenings + weekends)
Dental practices	32–42%
Medical primary care	28–38%
Medical specialty	30–40%
Behavioral health	38–48% (heavy evening tilt)
Legal initial consultations	35–45% (heavy weekend tilt)
Hotels	50–65% (24/7 by category)
Salons / spas	25–35% (evening peak)
Veterinary	30–40% + emergency surge
HVAC / plumbing / electricians	35–55% (emergency-heavy)
Restaurants (reservation lines)	40–55%

The numbers shouldn't be surprising once you think about it: customers think about their dental appointment when they're brushing their teeth at 10pm. They check on a hotel reservation when they're packing Sunday morning. They call about an HVAC issue when the heater fails Saturday night.

The historical SMB response: voicemail, an answering service, or nothing. Result: most of that demand walks.

The voicemail-to-booking conversion (and why it's so bad)

Walk through what happens to a missed after-hours call:

Step	Probability
Customer leaves a voicemail	30–50%
Voicemail returned the next business day	50–75% of those left
Customer answers the callback	40–60% of returned
Callback converts to a booking	30–50% of answered
Net: after-hours call → booking	2–10%

So of 100 after-hours calls that staff couldn't answer, 2–10 eventually become bookings. The other 90–98 are gone.

The destruction is not because the customer doesn't want to book. It's because the business asked them to wait, leave a message, and remain interested until tomorrow — a level of patience customers in 2026 don't have.

The AI front desk conversion (and why it's so much better)

The same 100 after-hours calls, routed to an AI front desk:

Step	Probability
AI answers within 1–2 rings	100%
AI correctly identifies intent	90–97%
AI books or escalates appropriately	85–95%
Booked calls actually show up (no-show rate same as in-hours)	~93%
Net: after-hours call → kept booking	35–55%

The 5–10× improvement comes from removing the wait. The customer who wanted to book at 9pm Tuesday books at 9pm Tuesday. No callback queue, no "I forgot," no "I went with the competitor who picked up."

The revenue math: a dental practice example

Dental practice doing 1,000 calls/month, 35% after-hours, baseline 5% after-hours → booking via voicemail callback:

Metric	Pre-AI	With AI front desk after-hours
Calls/month	1,000	1,000
After-hours calls	350	350
After-hours conversion to booking	5% (~17 bookings)	42% (~147 bookings)
Incremental after-hours bookings	—	+130/month
Avg new-patient first-visit value	$325	$325
Avg LTV new patient	$4,500	$4,500
First-visit revenue / month from incremental	—	+$42,250
LTV-adjusted incremental value (40% new-patient share)	—	+$234,000/month value created
AI front desk cost (voice + text)	—	$700/month
Net first-month impact	—	+$41,500 (cash terms)

The LTV-adjusted number is the right way to think about it long-term, but even the cash-terms first-month impact is 50–60× the cost. After-hours is the single highest-leverage deployment point for most SMBs.

Per-vertical playbooks

Dental and medical practices

Evening flow (5pm–11pm):

• Routine bookings, reschedules, FAQ
• Patient questions about morning visits ("can I eat before my procedure?")
• Insurance / billing questions

Weekend flow:

• New-patient inquiries (the biggest weekend bucket)
• Same-week and next-week bookings
• Provider research (Google reviews mention practice, customer wants to book)

Emergency policy:

• Define clearly what's emergency vs. urgent (pain level, infection signs, trauma)
• AI provides triage guidance per practice policy + immediately escalates to on-call
• Never delay emergency response to attempt scheduling

Hotels and hospitality

24/7 by definition — already operating round-the-clock; AI replaces or supplements night-shift front desk staff.

• Pre-arrival questions (parking, check-in time, late arrival)
• Local recommendations
• In-stay requests (room temperature, extra towels, room service)
• Post-stay billing questions and review prompts

Legal initial consultations

Weekend tilt heavy — many people research lawyers on Saturday mornings.

• Intake questions and conflict-check pre-flight (covered in law firm post)
• Initial-consultation booking with structured intake form
• Empathetic handling of distressed callers (DUI, domestic, criminal at 2am — these go to humans immediately)

HVAC / plumbing / electricians

Emergency-heavy after hours.

• Service intake with urgency triage (no water, no heat, sparking outlet → emergency dispatch)
• Same-day vs. next-business-day scheduling
• Up-front pricing or visit-fee disclosure

For these verticals, the AI's job is mostly to capture intent, dispatch emergency calls, and schedule non-urgent for next business day.

Salons and spas

Evening peak for next-day and weekend bookings.

• Book + provider preference + service add-on
• Cancellation and reschedule
• Often combined with upsell at booking (upsell playbook)

Veterinary

Day-of-emergency + weekend — pet medical concerns spike weekends.

• Triage (is this an emergency? routing to ER vs. next-day appointment)
• Routine booking for non-urgent
• Empathetic handling — pet anxiety is real customer anxiety

The 24/7 deployment specifics

Three engineering details to get right for a 24/7-grade deployment:

1. The phone system handoff

Most SMBs use a VoIP provider (RingCentral, GoTo, Vonage). Configure:

• Business hours: ring to staff first (3–4 rings), no-answer-forward to AI
• After business hours: ring directly to AI, with a brief greeting that mentions current hours
• Holiday calendar: maintained centrally, AI handles holiday calls and references next open day

Test the configuration with test calls at multiple times of day.

2. Calendar awareness

The AI needs to know:

• Which providers / rooms are available on which days
• Which slots are bookable by AI vs. staff-only (e.g., new-patient consults might require staff verification)
• Same-day vs. next-day vs. emergency slot rules
• Provider time-off and out-of-office calendars

Calendar integration matters more for 24/7 than for in-hours coverage because there's no one to override the AI's decision in real time.

3. Emergency escalation paths

Each vertical has its own emergency definition. The AI front desk must:

• Recognize emergency language across multiple phrasings
• Read a clear emergency response per practice policy (e.g., "go to nearest ER," "call 911," "the on-call provider will call you within 15 minutes")
• Page the on-call human with full transcript and caller info
• Continue to handle the conversation while the human is being paged

The escalation path is on-call rotation, paging system (PagerDuty, OpsGenie), and human-acknowledgment requirement. Without this layer wired correctly, after-hours emergency handling is the AI's biggest liability.

The customer experience at 10pm

What it actually feels like for a customer calling at 10pm Tuesday:

Ring once.

AI: "Hi, thanks for calling [Practice]. I'm here to help — what can I do for you?"

Customer: "I'm trying to book a cleaning."

AI: "Sure, I can help with that. Are you an existing patient or new to the practice?"

Customer: "Existing — I'm Maria Chen."

AI: "Great, hi Maria — let me pull up your file. I see your last cleaning was 5 months ago, so you're due. I have openings next Monday at 8:30am, Tuesday at 1pm, or Friday at 4pm. Any of those work?"

Customer: "Tuesday at 1pm, perfect."

AI: "Booked — Tuesday June 4th at 1pm with Linda. You'll get a text confirmation. Anything else?"

Customer: "Actually, can I add a whitening?"

AI: "Absolutely — we offer in-office whitening for $189. I'll add 30 minutes to your appointment and update your time to 12:30pm. Sound good?"

Customer: "Yes thanks."

AI: "Done. See you Tuesday at 12:30pm. Have a good evening!"

Total time: 90 seconds. Customer experience: better than calling at 10am Wednesday and waiting on hold.

The honest limits — when humans still win

Three scenarios where the AI should not handle after-hours, even when capable:

1. Distressed callers. A divorce inquiry at 1am, a behavioral-health crisis call, a fraud-victim call to a financial advisor — these need a human, immediately, with empathy. The AI's job is to recognize the situation, hand off as fast as possible, and provide interim support.

2. VIP relationships. High-value, long-relationship customers expect to talk to a human even after hours. Tag these accounts and route directly to on-call.

3. Complex transactions. A hotel guest disputing a $4,000 corporate charge at 2am gets a human. The AI's role is to escalate and reassure ("I'm getting the night manager for you right now").

Where Twig fits

For SMBs, Twig handles text-side 24/7 coverage:

• Website chat at 2am — visitor asking about services, hours, pricing, booking
• In-app help for SaaS — password resets, billing questions, support intake
• Inbound email outside business hours — info@, hello@, support@ triaged and answered immediately
• SMS conversations triggered by missed calls or web visits

Voice-channel 24/7 pairs with a voice AI front desk vendor. Same shared substrate — calendar, CRM, knowledge base, escalation paths.

The bottom line

After-hours demand is real, measurable, and currently mostly lost. SMBs that capture it via AI front desk see immediate pure-incremental revenue — no cannibalization of in-hours, no impact on staff workload, just bookings that wouldn't have existed otherwise. The math at scale is compelling enough that for most SMBs the after-hours case alone justifies the AI front desk deployment, and every other use case (no-show reduction, copay capture, upsell, missed-call recovery) is found money on top.

The first 30 days of a 24/7 deployment usually reveal a surprising number: the practice was missing more demand than the owners realized. Once captured, the question isn't whether to keep the deployment — it's how to expand it.