This is a snapshot of Sunday Roast's own past event records. It is not a market-wide Singapore catering benchmark and should not be read as one.
The hosted records run from 12 January 2017 through 24 April 2026. They show the scale and mix of Sunday Roast's recorded history, with the gaps disclosed instead of estimated.
Sunday Roast's current operating model is also simple: food is supplied raw for cooking onsite. The host or venue team can cook, or chef-led grilling can be attached to the food order. Historical internal service labels are not used in this analysis.
What this snapshot covers
A first-party record summary with a fixed date range and visible exclusions.
The population is Sunday Roast's own past-dated hosted-event archive through 24 April 2026. Pending, postponed, cancelled, soft, and future-dated rows are not counted as hosted events.
These figures come from Sunday Roast's own event records, not a survey of Singapore hosts or caterers. They show the events in Sunday Roast's history, but they do not establish market demand, industry averages, or another caterer's event mix.
Only aggregates are published here. Customer contact details, addresses, spend, and private row-level booking data remain outside this public snapshot.
The headline Sunday Roast records
The normalized public snapshot records 1,290 hosted events since 2017 and 21,274 recorded guests. It also identifies 188 returning hosts and 24 companies approved for public proof use.
These are booking counts, not ratings or satisfaction claims. A hosted event means one qualifying booking record exists; a returning host means the same host can be linked to at least two hosted events.
Why the guest-count denominator is 1,001, not 1,290
A positive pax value was detected for 1,001 of the 1,290 hosted records. That is 77.6% coverage. The remaining 289 hosted records have no detected positive pax value.
Those missing values are not guessed or filled. Every guest-count percentage below therefore uses 1,001 records as its denominator.
The approved guest total across all records is 21,274. The group-size breakdown below uses only the 1,001 records with a recorded guest count above zero.
How the recorded guest counts are distributed
Percentages below use only the 1,001 hosted records with detected positive pax.
The largest band is 11–20 guests: 509 records, or 50.8% of the recorded-pax set. The next largest is 21–30 guests: 207 records, or 20.7%.
Taken together, 716 records sit between 11 and 30 guests, equal to 71.5% of the recorded-pax set. A total of 875 records, or 87.4%, are for 30 guests or fewer.
The 51+ band is deliberately combined. Publishing broader aggregate bands makes the result useful while reducing the chance that a large private event could be identified from a small cell.
- 1–10 guests: 159 records (15.9%)
- 11–20 guests: 509 records (50.8%)
- 21–30 guests: 207 records (20.7%)
- 31–50 guests: 99 records (9.9%)
- 51+ guests: 27 records (2.7%)
Private records form most of the hosted archive
Of 1,290 hosted records, 921 are classified as private events and 369 as company or office events. That is a 71.4% private and 28.6% company-or-office split.
This is a classification of Sunday Roast's archive, not a statement about the broader Singapore BBQ catering market. Classification quality also depends on the normalized signals preserved in the historical records.
What the returning-host record proves
The normalized archive contains 984 contact records linked to at least one hosted event. Of those, 188 are linked to at least two hosted events, equal to 19.1% of hosted contact records.
That supports a repeat-booking count. It does not support a claim about customer satisfaction or explain why anyone returned. A quotation, survey response, or other approved evidence would be needed for that story.
The same rule applies to individual repeat-record cards: the site can state the confirmed booking count while keeping inferred motives and outcomes out of the copy.
Methodology and interpretation limits
The method keeps every published total repeatable and makes its limits visible.
The source pipeline starts with 1,389 recorded event rows. Normalization produces 1,296 canonical unique events after 93 duplicate rows are collapsed. Applying the hosted-status and date rules leaves 1,290 hosted records.
The guest-count analysis includes only records with a recorded guest count above zero.
Returning hosts are counted when the same host can be linked to at least two hosted events.
Company and office totals combine records classified as corporate or office; private totals use the private classification.
A deterministic generator checks these figures against the committed metric snapshot and public proof contract. If the inputs change, its check fails until the public benchmark JSON is rebuilt and reviewed.
- Do not infer customer satisfaction or reasons for repeat business.
- Do not infer seasonality, lead time, venue frequency, menu frequency, or historical pricing.
- Do not use historical internal service labels to describe current fulfilment.
- Do not present this first-party snapshot as a Singapore industry benchmark.
- Do not combine aggregate repeat-record guest counts as if they came from one event.