Marlowe & Finch Front of house · Guests
Guest preferences & loyalty

Remember the guest, not just the booking.

A quiet memory for a 60-seat room — who likes which table, what they drink, and what they can't eat — surfaced to the floor at the moment it matters, and gated by what each guest has actually agreed to. Read by Priya Sharma, Floor Manager

Returning guests on file
214
Two or more visits, identity-resolved across phone & bookings
Average visits per guest
3.4
Across the returning-guest book
Consented for marketing
61%
The other 39% are served, never messaged
Allergies flagged
38
Surfaced to the floor for every booking, consent or not

One guest, remembered

A light CRM record — assembled from bookings, the phone line and Tommo's floor notes.

Priya Tan

+65 8421 7733  ·  First visit Dec 2024  ·  Usually a party of 2
6 visits in 18 months
Allergy · floor Tree-nut allergy. Flag to kitchen and section on seating. Keep her clear of the smoked ocean trout (bunya nut), the charred broccolini (macadamia) and the Manjimup truffle ice cream (hazelnut). This note shows on every booking — it is never gated by marketing consent.
Observed preferences
  • Usually the Manuka-smoked duck with the Bridge Road porter. Tommo's pairing — she's ordered it four of six visits. From POS & floor notes
  • A table away from the pass. Finds the room loud; happiest in the window two-top. Noted by Priya, visits 2 & 4
  • Finishes with the Rutherglen Muscat, not coffee. From POS
Visit & spend history
  • Dec 2024First visit · two$148
  • Mar 2025Dinner · two$172
  • Jun 2025Dinner · two$166
  • Sep 2025Birthday · four$310
  • Feb 2026Dinner · two$184
  • Apr 2026Dinner · two$196

Spend per head trending up — roughly $74 at first visit to $98 now, in line with the repeat-guest pattern.


For tonight's 19:30 booking

A prompt for the floor, not an instruction to a machine. Priya decides what happens.
1

Seat her in the window two-top, away from the pass

She finds the room loud and has asked for it before. If it's spoken for, the corner banquette is the next-quietest.

2

Have the Bridge Road porter ready for Tommo's table visit

Four visits in a row it's been the duck and the porter. Worth a quiet word — and an alternative if she fancies a change tonight.

3

A quiet "happy returns" — her sixth visit

No fuss and no badge on the table. Just so the floor knows she's a regular, and Tommo can say so if the moment's right.

!

Tree-nut allergy — carried through to the kitchen ticket

Steer her clear of the ocean trout, the broccolini and the truffle ice cream. This rides on the booking regardless of anything else.

Nothing here happens by itself. The system surfaces the few facts that matter at booking and at arrival; the porter, the table and the word at the table are Priya's and Tommo's call. The allergy line is the one thing it will not let a service forget.


What we do — and don't do — with a guest's data

PDPA in practice: consent gates marketing; safety does not wait on consent.
Autumn tasting-menu campaign · audienceMarketing list
Priya Tan6 visits · consent given May 2025 Included →
Marcus Lee3 visits · consent given Jan 2026 Included →
Wei Goh4 visits · marketing consent not given Excluded — not messaged
Aisha Rahman2 visits · consent given Mar 2026 Included →
Daniel Ong5 visits · withdrew consent Apr 2026 Excluded — opted out

Wei Goh and Daniel Ong are among our most frequent guests — and they are not on this list. A guest who has not opted in, or who has opted out, is simply not contacted. Their table is still set exactly as they like it.

The rules, plainly
  • Consent gates marketingOnly guests who have opted in are messaged. Withdrawing consent removes a guest from every campaign — no exceptions, no re-asking at the door.
  • Purpose limitationObserved preferences — her table, her porter, her muscat — are used to serve her on the night. They are not sold, shared, or repurposed for anything she didn't agree to.
  • Access & correctionA guest can ask to see what's on file and have it corrected or removed. We treat that as a feature of the relationship, not an awkward request.
  • Safety is not gatedA tree-nut or shellfish allergy is force-surfaced to the floor for every booking, whatever a guest's marketing choice. Safety never waits on a consent flag.
Under the hood What this is today — and where the guest memory should go next

Today this is a thin guest-profile store keyed by phone number, holding visit history, observed preferences, allergies and a marketing-consent flag — with retrieval at two moments only: when a booking is taken, and when the guest arrives. The profile on this page is illustrative, built from synthetic seed data, not a live customer record. The honest point of the demo is not the datastore — Singapore is well served by local platforms for that — but the PDPA consent and safety layer that sits on top of it.

On the roadmap

  1. Buy the store, don't build it. A real 60-seat room should sit this on an SG-native platform — Advocado or Oddle for cost, Eber or SevenRooms for deep reservation-linked profiles. The value we add is the compliant retrieval and consent governance, not another CRM.
  2. Identity resolution across channels. Merge the same guest seen on the phone line, the web booking form and the POS into one record — carefully, since a wrong merge mixes up two people's allergies. Conservative matching with a human check beats an over-eager auto-merge.
  3. Self-service access & correction. Expose the PDPA access/correction right as a guest-facing feature — a guest can see and edit what's held, and withdraw consent in one tap, with the change reflected immediately across every campaign audience.
  4. Surface at arrival, not just booking. A quiet host-tablet ping when a known guest is seated, showing the same few facts — visits, seat, allergy, last note — so the floor doesn't rely on memory across 60 covers and a changing roster.
  5. The guardrails don't move. Marketing stays consent-gated and purpose-limited; allergies stay force-surfaced regardless of consent; preferences serve the guest and are never sold. The system proposes a personal touch — a person always delivers it.
Full use-case write-up — the problem, benefits and vendor landscape: UC06 — Guest preference & loyalty → Concept mock-up — illustrative data, not a live system.