AI in Singapore F&B: Customer preference / loyalty memory
Last updated: 2026-05 FX reference: 1 USD = SGD 1.28, retrieved 2026-05-28
See the mock-up for this use case
A static, on-brand design concept — illustrative data, not a live system.
The problem this solves
A 60-seat owner-operator has no reliable way to recall "Mrs Tan, fourth visit, prefers the window, allergic to shellfish." Repeat guests spend about 67% more per order than first-timers, and fine dining draws roughly 51% of sales from repeat customers (National Restaurant Association via Restroworks, independent). Without a memory, every visit starts from zero.
Retention is the highest-leverage lever when margins are thin. With SG F&B net margins at 5–7% and over 2,400 closures in ten months of 2025, a 5% lift in retention raises profits 25–95% (Bain/Reichheld, HBR — academic).
Recalling allergies and preferences is also a safety and service issue: remembering a shellfish allergy across 60 covers and several staff is unreliable by hand. A system that surfaces the right note at booking and arrival fixes both the service and the safety gap — provided it handles personal data correctly.
What it costs to ignore
No credible standalone SGD figure. The best proxies are the foregone 25–95% profit upside from a 5% retention gain (Bain, academic) and the +67% repeat-guest spend gap. For a 60-seat venue at SGD 80 average spend, converting 50 lapsed guests a year to one extra visit each is about SGD 4,000/year — illustrative.
What good looks like
- Higher retention — realistic floor: nudging lapsed guests recovers a measurable share of repeat visits; the conservative anchor is Bain's 5% retention → 25% profit.
- Higher per-visit spend — recalled preferences enable relevant upsell (+67% repeat-spend gap, independent).
- Service and allergen safety — the right allergy and seating notes reach every staff member.
- First-party data ownership — you hold the guest relationship rather than the aggregators.
How it works
A guest profile is assembled from scattered data and shown at booking and arrival.
- Mechanism: a profile for each guest, keyed by phone or email and built from several sources, is retrieved at booking and at arrival, with consent controlling what is stored and shown.
- Data it draws on: booking history, phone enquiries, review mentions, spend, and stated preferences and allergies.
- How it decides: it links records for the same guest across channels and surfaces the few most useful facts — visit count, seating preference, allergies, last issue — at the right moment, within PDPA limits.
Typically built on a Singapore loyalty/CRM platform (such as Eber or Advocado) with a retrieval step at booking and arrival.
Vendor landscape
Singapore-native gap. This is the use case where SG-native options clearly exist. Eber and Advocado are genuine Singapore CRM/loyalty platforms, alongside Oddle loyalty and Qashier's built-in "Treats." The market is well served locally — the question is which tier fits, and how you handle consent and data, not whether to build one.
| Vendor | Origin | SGD/month | Suited to |
|---|---|---|---|
| Advocado | Singapore | SGD 200/outlet + ~SGD 1.35/enrol | SG SMEs wanting local, SGD-billed CRM |
| Eber | Singapore | Lite ≈ SGD 114, Standard ≈ SGD 383 | Feature-rich SG loyalty at scale |
| Oddle | Singapore | SGD 0 + ~SGD 0.35/enrol | Operators already on Oddle ordering |
| Qashier (Treats) | Singapore | Free with Essential POS | Operators running Qashier POS |
| SevenRooms | US (SG office) | from ≈ SGD 639, annual | Fine-dining reservation-linked recall |
A few facts to weigh: Eber's pricing is now USD-denominated (Lite USD 89 ≈ SGD 114) and rises steeply between tiers, with the entry tier capped at 5,000 contacts and one store. Advocado prices per outlet plus a per-customer enrolment fee. Oddle has the lowest entry cost but thinner CRM depth. SevenRooms has a real Singapore footprint and suits fine dining, but is annual-contract and quote-only. Enterprise loyalty platforms (Punchh, Paytronix) are over-scoped for an SME.
Buy or build?
Buy an SG-native platform — Advocado or Oddle for cost, Eber or SevenRooms for deep reservation-linked profiles. Building a compliant customer-data platform isn't justifiable at 60 seats. The work worth doing yourself is the consent and retrieval workflow on top: getting the right note to the right staff member at the right moment, with the guest's permission.
Singapore-specific considerations
- PDPA (this is the PDPA-heavy use case): get consent for the stated purpose; don't make it a forced condition of dining beyond what's reasonable; don't repurpose the data without fresh consent; let guests access and correct their profile; notify the purpose at collection.
- Grants: PSG (70%, enhanced April 2026, SGD 30,000/year cap) for a pre-approved loyalty SKU; AI Singapore 100E (up to SGD 150,000) for an AI personalisation build.
- Integrations: Eber ↔ SevenRooms/Oddle/Eats365; Advocado ↔ Qashier/Eats365.
- Language: guest comms should support English and Mandarin at minimum, with Bahasa Melayu and Tamil for inclusivity.
Sources
- Restroworks — customer retention statistics (NRA data) — https://www.restroworks.com/blog/customer-retention-statistics-restaurants — independent — accessed 2026-05-28
- HBR — value of keeping the right customers (Bain/Reichheld) — https://hbr.org/2014/10/the-value-of-keeping-the-right-customers — academic — accessed 2026-05-28
- Eber pricing — https://eber.co/pricing — vendor (SG) — accessed 2026-05-28
- Advocado pricing — https://advocadoapp.com/pricing-sg — vendor (SG) — accessed 2026-05-28
- Oddle SG — https://www.oddle.me/sg — vendor (SG) — accessed 2026-05-28
- PDPC — data protection obligations — https://www.pdpc.gov.sg/ — regulator — accessed 2026-05-28
This is part of a series on AI use cases for Singapore F&B operators, refreshed every two months. If you'd like to discuss applying any of this to your restaurant, get in touch at [email protected].