The evidence covers about 13 months across 2023 and 2024
The sample runs from 2023-07-17 to 2024-08-20. It supports diagnosis of this defined period, not claims about the store’s full operating history.
Using 25 complete, unique low-rating review records spanning about 13 months, this report identifies concentrated operating signals without extending the evidence beyond the audited sample.
The sample runs from 2023-07-17 to 2024-08-20. It supports diagnosis of this defined period, not claims about the store’s full operating history.
The 25 records contain 20 one-star and five two-star reviews, with no unrated records. The rated-sample average is 1.20.
All 25 App-tagged records are present, unique and complete. Duplicate count, empty-body count, truncation count and App gap are all zero.
Counts describe the audited low-rating records only. They do not represent total review volume or the store’s overall negative-review rate.
Two sample stages—H2 2023 and 2024—cover the approximately 13-month window. Every stage metric is recalculated from review-level tags.
Use one operating control loop across the full guest journey: ingredient acceptance, pre-service inspection, table ownership, timed serving, transparent pricing and manager closure of every exception.
A review may match multiple semantic categories. Positive statements are excluded from negative issue tags, and category totals must not be added together.
Darker cells indicate more review-level matches in that year.
Product groups are assigned review by review and aggregated directly from the 25 source records.
Record ingredient acceptance, thawing and holding time; taste-test soup bases; verify refill-station availability; and time meat, drinks, seafood, staples and snacks separately.
P0 contains safety, hygiene and service-control failures. P1 restores product consistency, speed, value and rule transparency.
Search all 25 source records by rating, year, issue or keyword. Faithful English translations are provided, and the original Chinese remains expandable in English mode.