Back to All Research
Plain-language summary of PRCA + Kano

What Hotel Guests Penalize and Reward Most in Vietnam Booking Reviews

From 97,014 reviews, this page shows which service issues hurt ratings most and which upgrades help only after basics are stable.

  • PeriodJan 2023 - Jan 2025
  • SourceBooking.com Vietnam
  • Sample97,014 reviews
  • ModelRegression with robust errors (R2 = 0.220)
Abstract Snapshot

Start here (new readers)

  • Penalty means score loss when an attribute fails. In this study, penalties are stronger than rewards.
  • Cleanliness is the top must-not-fail factor.
  • Staff service is the strongest upside lever when executed well (+0.680 reward coefficient).
  • Attribute types: 2 Basics, 4 Performance drivers, 2 Delighters.

How to read terms: Basic = guests expect it; Performance = better service raises scores directly; Delighter = bonus value after basics are stable.

Source note: all coefficients shown are reported outputs from viz-prca-mirrored-effects.svg.

Authors: Chung Hy Dai, Anh Minh Hoang

Contact: hmianh2504@gmail.com

Core Metrics

  • Sample size

    97,014

  • Approx. hotels

    ~1,000

  • Average guest score

    8.83 / 10

  • Share of perfect 10 scores

    49.2%

  • Model explanatory power

    22.0% (R2 = 0.220)

  • Uncertainty check

    Robust standard errors (HC1)

  • Attribute type split

    2 Basics / 4 Performance / 2 Delighters

  • Strongest upside driver

    Staff +0.680

Visual Context

Hotel touchpoints (illustrative)

These example visuals help non-technical readers connect the model outputs to real guest touchpoints. They are illustrative, not measurement data.

Signature boutique lobby with layered pendant lights and lounge seating
Arrival experience: first impressions and emotional framing.
Upscale hotel room with warm lighting and neutral textile layering
Room quality cues reinforce performance attributes.
Resort pool deck at golden hour with open skyline
Facilities and leisure amenities function as delighters.
Contemporary boutique suite with curated lighting and minimalist decor
Premium suite storytelling supports pricing and value perception.
Visualization

Two Charts You Need

Read left first (impact size), then right (attribute type and fix order).

Mirrored penalty and reward coefficients for eight hotel attributes
Penalty (left) and reward (right) coefficients for each attribute. Larger absolute value means stronger influence on review score.
Asymmetry class map for Basic, Performance, and Excitement attributes
Attribute type map used for sequencing: Basics first, Performance second, Delighters last.
Method in Plain Steps

How the analysis was built

Data collection

Collected Booking.com reviews and removed duplicates, resulting in 97,014 usable records.

Language standardization

Used NLLB-200 to translate and standardize multilingual reviews into one analysis-ready format.

Attribute mapping

Used Sentence-BERT to map text into 8 service attributes (reported proxy precision up to 0.92).

Polarity construction

Separated positive and negative statements to estimate gains and losses independently.

PRCA + Kano classification

Estimated penalty/reward effects with robust regression, then grouped attributes into 2 Basics, 4 Performance drivers, and 2 Delighters.

How to Use Both Studies

How the segmentation page and this PRCA page connect

Recommended order for newcomers: identify who is struggling first, then choose what to fix first, then set budget sequence.

Step A. Segmentation page

  • 74,896 reviews across 458 Da Nang hotels.
  • Shows which guest segments face the most friction.
  • Helps decide which audience to prioritize first.
Open Segmentation Study

Step B. PRCA page (this page)

  • 97,014 reviews across approximately 1,000 properties.
  • Quantifies how much each of 8 attributes can hurt or help scores.
  • Classifies attributes into 2 Basics / 4 Performance / 2 Delighters.

Use this layer to decide where quality failures must be prevented first.

Step C. Connected execution

Pick target segment Rank attributes by penalty Sequence operations and budget

Segmentation picks the right guest target; PRCA ranks the fix order for that target.

Decision Rules

Practical Fix Order for Managers

Protect Basics

Stabilize high-penalty factors first.

Optimize Performance

Improve staff, room, value, and location next.

Differentiate with Delighters

Invest in facilities and food last.

Coefficient format: penalty / reward. More negative penalty means stronger score damage when that attribute fails.

  • Cleanliness-1.500 / +0.145
  • WiFi-0.656 / -0.052
  • Staff service-0.965 / +0.680
  • Room quality-0.967 / +0.222
  • Value-0.382 / +0.090
  • Location-0.341 / +0.034
  • Facilities-0.115 / +0.410
  • Food-0.158 / +0.525

Execution Rhythm

Pass 1

Set and enforce minimum standards for cleanliness and WiFi.

Pass 2

Standardize frontline service and room consistency.

Pass 3

Roll out selective facility and food upgrades after basics stabilize.

Review monthly, rerank quarterly, and keep premium upgrades gated on stable basic performance.

Back to All Research