The problem
A 3-restaurant group in South Mumbai (modern cuisine, 40–60 covers per venue, average check ₹2,200) was losing bookings in three distinct ways:
- After-hours blackout: the reservation phone was unmanned 11pm–2am — exactly when their Instagram ads converted. A weekday night brunch ad at midnight generated enquiries that went to voicemail.
- Hinglish miscommunication on calls: staff trained in English struggled with Hindi-speaking regulars booking for "chaar log, aath baje" — slot and party size got entered wrong 1 in 20 times.
- 18% no-show rate: Saturday night loses hurt most. A 50-cover restaurant with 18% no-shows is 9 empty seats in prime time = ~₹20k walked out the door per Saturday.
What we built
An AI voice bot on their shared reservation number. The bot:
- Speaks Hinglish natively — handles "Mujhe Saturday ko dinner chahiye, 4 log, 8:30 baje" and "I need a table for 6 at 7pm" in the same call flow
- Reads live cover availability across all 3 restaurants + their lounge area; offers alternatives when first choice is full ("Colaba is full at 8, but 8:45 works, or Bandra at 8 sharp — which?")
- Books into their existing reservation system via webhook (no new software for the host to learn)
- Fires a WhatsApp confirmation within 60 seconds with Google Maps pin, dress code reminder, happy-hour window, and a one-tap cancel link
- Sends a 2-hour-before reminder with one-tap "Confirm" or "Reschedule" — the core of the no-show reduction
- Pings the on-duty manager via WhatsApp if a VIP guest (flagged in CRM) books
The numbers (first 60 days)
| Metric | Before | After |
|---|---|---|
| Reservation line answer rate (11pm-2am) | 0% | 100% |
| Reservations captured after hours | ~0 | 180/month |
| Avg booking party size via bot | — | 3.4 people |
| Average check × party = revenue/booking | — | ₹7,480 |
| After-hours revenue captured | ₹0 | ₹2.4L/month* |
| No-show rate (with WhatsApp 2-hr reminder) | 18% | 7% |
| Saturday prime-time cover utilisation | 82% | 96% |
| Host hours spent on phone | ~28 hr/week | ~9 hr/week |
* Revenue capture = bookings × avg check × (1 − no-show rate). Assumes 80% of bot-booked covers would otherwise have been lost entirely.
Economics
At ₹1.99/minute with average call duration 2.8 minutes, roughly 350 bot-handled calls/month = ~₹1,950/month in bot airtime. WhatsApp confirmation + reminder messages (720/month at ₹2/msg via Akashvanni) = ~₹1,440/month. Combined monthly bot + messaging cost: ₹3,390.
Captured revenue: ₹2.4L/month in after-hours bookings + ~₹44k/month in no-show recovery (reduction × avg check × cover count). ROI: ~70× on monthly running cost.
Tech stack
- Voice layer: Appzur Voice AI with Hinglish code-switching (trained on Mumbai accents)
- Reservation backend: webhook integration with their existing booking system (we don't force a vendor change)
- Messaging: Akashvanni (Appzur's WhatsApp platform) for confirmations + reminders + VIP alerts
- CRM tagging: regulars + VIPs flagged via phone-number match
- Timezone awareness: IST-locked (international diners booking from US/UK time zones auto-converted)
Setup
Day 1 morning — brief call with the owner + head host. Captured the booking flow, VIP list structure, reservation-system API docs, dress-code & happy-hour rules per restaurant.
Day 1 afternoon — bot trained on their 3 menus, timing rules, and typical Hinglish phrases their regulars use. WhatsApp templates (4 total) submitted to Meta.
Day 2 — WhatsApp templates approved. Soft-launch on one restaurant for 24 hours, owner reviewed every recording. 3 wording tweaks applied.
Day 2 evening — live across all 3 restaurants. Total setup: 48 hours calendar time, ~6 hours of owner's time.
"The no-show number is the one that made me a believer. I've been trying to crack that for 8 years. A ₹2/message WhatsApp reminder did what deposits and penalty clauses never could."
What this means for your restaurant
This pattern works wherever (a) more than 30% of reservations come via phone, (b) the no-show rate is above 10%, and (c) you have at least one location in a market where WhatsApp penetration is >80% (basically all of urban India). It does NOT work well for walk-in-only concepts or food courts. For delivery-first or cloud kitchens, pair the voice bot with a WhatsApp ordering flow — a different build, happy to scope it.
Want this for your restaurant?
₹1.99/minute. ₹2/WhatsApp message. ₹999 setup. 48-hour go-live. No monthly subscription.
Book a 15-min scoping call →