Standing benchmark · updated annually
Southeast Asia F&B Retention Benchmark 2026
Updated August 2026 — Most Southeast Asian F&B venues cannot state their own repeat rate, which is why retention decisions get made on instinct. This benchmark publishes one comparable set of retention measures per market, with the method fixed in advance and every figure sourced from anonymised, consented venue data.
A standing annual benchmark of repeat rate, 90-day lapse rate, enrolment capture rate and win-back response across Vietnam, Malaysia, Singapore and the Philippines. Methodology and metric definitions are final and published below. The 2026 collection window closes on 2026-12-31 and the figures publish in January 2027 — until then each measure shows its reporting state rather than a stand-in number.
Methodology
- Population
- Independent and small-group F&B venues (1–10 outlets) in Vietnam, Malaysia, Singapore and the Philippines.
- Sample size
- Target of at least 1,200 qualifying venues across the four markets, with a floor of 200 venues per market so no market split is published on a base too small to report. The final achieved n, per market, publishes with the figures in January 2027.
- Data source
- Anonymised, aggregated transaction and membership data from consenting venues, plus a structured operator survey for venues without a connected system.
- Identity resolution
- A guest is one resolved member identity (phone or messaging ID). Anonymous walk-in transactions are excluded from repeat-rate denominators.
- Reporting period
- Rolling 12 months to the end of the collection window; comparisons are like-for-like venues only.
- Exclusions
- Venues open under 6 months, and venues with under 200 identified transactions in the period.
- Weighting
- Reported unweighted by venue, with market splits shown separately rather than blended into one regional average.
- Margin of error
- Reported per market as a 95% confidence interval on each rate, computed from the achieved venue base rather than assumed. Any market split whose interval is wider than ±5 percentage points is published with that interval shown next to the figure instead of being quoted as a point estimate.
- Conflict of interest
- PEKO builds retention software and therefore has an interest in these numbers. The method is fixed before collection, venue-level data is never published, and the raw definitions here are the ones used in every chart.
- Next update
- Annually, with the collection window and any method change noted in the changelog.
Anyone may reuse these figures with attribution to PEKO and a link to this page. Corrections and challenges to the method are welcome — the method is published before the numbers on purpose.
The measures
Four charts, one per metric, broken out by market. Bars render when the 2026 figures publish in 2027 · 01; until then each row states its reporting month rather than an illustrative number, because this page is built to be cited.
Repeat rate
Share of identified guests in the period who transacted two or more times at the same venue.
Unit: percent of identified guests
- Vietnam Reporting 2026-12
- Malaysia Reporting 2026-12
- Singapore Reporting 2026-12
- Philippines Reporting 2026-12
Target base: 1,200+ venues (min. 200 per market) · Collection window: 2026-01-01/2026-12-31 · Publishes 2027-01
90-day lapse rate
Share of previously active members with no transaction in the trailing 90 days.
Unit: percent of active members
- Vietnam Reporting 2026-12
- Malaysia Reporting 2026-12
- Singapore Reporting 2026-12
- Philippines Reporting 2026-12
Target base: 1,200+ venues (min. 200 per market) · Collection window: 2026-01-01/2026-12-31 · Publishes 2027-01
Enrolment capture rate
Share of transactions that result in an identified member, measuring how much of the door is actually captured.
Unit: percent of transactions
- Vietnam Reporting 2026-12
- Malaysia Reporting 2026-12
- Singapore Reporting 2026-12
- Philippines Reporting 2026-12
Target base: 1,200+ venues (min. 200 per market) · Collection window: 2026-01-01/2026-12-31 · Publishes 2027-01
Win-back response rate
Share of lapsed members who transact within 14 days of receiving a win-back message.
Unit: percent of messaged members
- Vietnam Reporting 2026-12
- Malaysia Reporting 2026-12
- Singapore Reporting 2026-12
- Philippines Reporting 2026-12
Target base: 1,200+ venues (min. 200 per market) · Collection window: 2026-01-01/2026-12-31 · Publishes 2027-01
Method changelog
- 2026-08-29 — Method, metric definitions, exclusions and weighting published in advance of collection.
- 2026-08-31 — Added target sample, margin-of-error method and the January 2027 release date.
FAQ
- When do the 2026 figures publish?
- January 2027, once the collection window closes on 2026-12-31. Until then each measure shows its reporting state rather than a number, because we publish the method before the results so the definitions can be challenged while they are still changeable — and so nobody cites a stand-in figure as a finding.
- Can I cite this benchmark?
- Yes. Reuse any published figure with attribution to PEKO and a link to this page. If you need the market split behind a chart, ask and we will send it.
- How do you define a repeat guest?
- One resolved member identity — phone number or messaging ID — with two or more transactions at the same venue inside the reporting period. Anonymous walk-ins are excluded from the denominator, which is why our repeat rates are not comparable to POS-reported figures that count receipts.
- Which markets are covered?
- Vietnam, Malaysia, Singapore and the Philippines, reported separately. We do not publish a single blended Southeast Asia number because the underlying channel and dining-frequency differences make it misleading.
- Is venue-level data ever published?
- No. Data is anonymised and aggregated, and no venue, brand or guest is identifiable in anything we release.
The PEKO ecosystem
PEKO and LOOP are two products from the same company. PEKO is the AI retention layer and runs alongside the POS you already use. LOOP is the AI-native POS that covers operations: recipe-level inventory, staff shifts, table plans and the kitchen display. Each works on its own, and run together they share one dataset, so nothing has to be entered twice. See PEKO + LOOP in one ecosystem