A night with Heep: 47 bookings saved between 7pm and 3am on a Friday
Eight hours of written messages, and the bookings that would have slipped through the cracks. The minute-by-minute log of a Friday night with Heep on the written channels.
6:59 pm, a Friday. The room is set, the brief just finished, the phone is already buzzing. Three calls on hold. Two Instagram DMs. One email in Italian.
That isn’t an organisation problem: six points of friction in 90 seconds, just as the first guests push the door open. Nobody has three hands.
I followed that night channel by channel.
47 Bookings saved over 8 hours on a typical Friday night. Without a single extra ring for your team.
6:59 pm: service starts, Heep is already on
Heep started at 5 pm. The written channels in parallel: WhatsApp, Instagram, Messenger, SMS, email, website form, plus follow-up on calls nobody could take. None waits its turn.
At 6:59 pm sharp, 14 requests already handled today.
The night hasn’t started.
7–9 pm: 1.4 requests a minute
Between 7 and 9 pm, 168 requests come in. One every 43 seconds.
No team mid-service holds that pace. That’s where 30% of the volume evaporates. At 8:15 pm your host is seating a table of eight, the pass is calling, the phone rings for the sixth time.
Of those 168 requests: 112 bookings, 38 modifications, 18 menu questions. All handled in under a minute by message. Mean for the night: 32 seconds.
- 168 Requests (in 2 hours, all channels)
- Under a minute From question to confirmation
- 0 Missed call left without a written reply (vs ~47 in a service without AI)
Want to see this on your volume? Simulate your Friday night
9:47 pm: the booking we saved
Case study
Table of 8, surprise birthday, £420 of revenue captured in 2 minutes
The guest calls at 9:47 pm. Room packed, nobody can take it. Heep sees the missed call and writes straight away: a table frees up at 10:15 pm, the slot is offered, a card pre-authorisation captured, a WhatsApp recap sent.
Nobody on the floor was pulled away; the host read the thread after the rush.
- £420 Revenue captured
- 2 min 14 to handle
- 0 Floor interruption
How Heep prioritises in real time
Not every message weighs the same. It’s the question I get most in training: what goes first? Four signals: intent, value, urgency, guest history.
Intent first
Booking, modification, complaint, question? Sorting happens before the greeting.
Then value
A table of 8 jumps ahead of a vegetarian menu question. Floor teams sign that off in two seconds.
Availability, never a double booking
Direct sync with Zenchef, SevenRooms or TheFork. Never two bookings on one slot.
Escalation, six precise cases
Groups over 10, private hires, complaints, critical allergies, identified VIPs, anything off-script. Your team takes over. On the rest, Heep closes the loop alone.
Before Heep / After Heep
Before
~47 missed calls over the night
Phones ringing on top of each other. Front-of-house yanked off the floor. 32% of volume never recovered the next day.
After
0 untreated request
Heep absorbs the peak, your team keeps the room. The 47 missed calls become 47 confirmed bookings.
11 pm–3 am: while you sleep
Service shuts at 11:30 pm. The kitchen is on the wash-up by 1 am. You switch off the phone.
During those four hours, 38 new requests come in. Jet-lagged tourists, Sunday brunch bookings, last-minute changes.
With nothing on the other side, those messages wait nine hours, and 35% go elsewhere. Here, the reply goes out in 41 seconds, in the guest’s language.
- 11:12 pm Table for 4, Sunday lunch, request from Munich (in German)
- 00:47 Shellfish allergy update on an existing reservation
- 02:33 Private hire asked for Thursday, passed to your team
Tomorrow starts with 12 confirmations in the book.
One thing to do at tomorrow’s open: ask your host how many calls they let ring last night. They’ll give a rough number. It will be below the real one.
Frequently asked questions
Can Heep really handle 1.4 requests a minute at peak?
Yes. The architecture is concurrent: each request runs in its own process. Peaks measured up to 4 requests a minute on real Moma Group venues, with no latency degradation.
What if Heep gets a high-stakes booking wrong?
Six precise cases trigger immediate human escalation: groups over 10, private hires, complaints, critical allergies, identified VIPs, anything off-script. On everything else, Heep closes the loop alone with a 94% confirmation rate.
Which channels are covered through the night?
WhatsApp, Instagram DM, Messenger, SMS, email and the website form, plus missed calls followed up by message. Everything lands in the same unified inbox for your team in the morning.
How long does it take to set Heep up?
A few days for a standard rollout. Sync with Zenchef, SevenRooms, TheFork or OpenTable. No data migration on your side.
