Customer Reviews & Sentiment¶
Dotlas aggregates customer reviews from diverse online sources including food delivery platforms, review aggregators, social channels, and brand pages. Each review is processed through advanced reasoning and natural language understanding models to generate:
- ABSA scores (Aspect-Based Sentiment Analysis) for
food
,service
,ambiance
,value
, and more - NPS Sentiment Rating (1–5 scale) aligned with Net Promoter Score methodology
- Categorized topics automatically extracted from the text (e.g.
staff friendliness
,food temperature
,accessibility
) - Time-series tracking of review quality per topic or outlet
This transforms qualitative customer voices into structured, quantifiable insights that can be compared across time, restaurants, brands, and topics.
Use Cases for Marketing, CX & Product Teams¶
- Track dNPS movement across time to detect brand health deterioration early
- Identify sentiment gaps — strong food but weak service? Value declining?
- Pinpoint CX strengths or issues at the brand, outlet, or city level using topic-tag trends
- Quantify qualitative feedback into structured, actionable dashboards
- Benchmark seasonal shifts in brand experience
- Use AI-tagged praise or complaints in marketing, operations and menu R&D
- Evaluate campaigns — do new activations lead to better NPS scores in target cities?
Data Dictionary — Individual Review Level¶
Each row represents one customer review as interpreted by the Dotlas AI pipeline.
Column Name | Data Type | Description |
---|---|---|
review_id | string |
Unique identifier for the individual review |
restaurant_id | string |
Dotlas-assigned ID of the restaurant being reviewed |
brand_id | string |
ID of the restaurant chain or brand |
review_text | string |
Raw text of the original customer review |
review_source | string |
Platform or website where the review was posted |
reviewed_at | timestamp |
Date and time the review was posted |
nps_sentiment_score | integer |
Customer sentiment on a 1–5 scale based on NPS reasoning |
aspect_food | float |
ABSA score for the food (1.0 to 5.0) |
aspect_service | float |
ABSA score for service (1.0 to 5.0) |
aspect_ambiance | float |
ABSA score for ambiance/environment |
aspect_value | float |
ABSA score for value for money |
topic_tags | array[string] |
List of detected topics (e.g. temperature , staff friendliness , portion size ) |
language | string |
Language of the original review |
neighbourhood_id | string |
Neighborhood ID of the restaurant reviewed |
city | string |
City where the review was recorded |
Data Dictionary — Brand-Level dNPS Trends¶
Each row aggregates customer review sentiment at the brand and time level for performance tracking.
Column Name | Data Type | Description |
---|---|---|
brand_id | string |
Dotlas-assigned brand identifier |
date | date |
Daily or weekly aggregation date |
city | string |
City where reviews were aggregated |
dNPS_score | float |
Derived Net Promoter Score (-100 to +100), based on NPS bucket distributions |
total_reviews | integer |
Count of reviews contributing to this aggregation |
avg_aspect_food | float |
Average food rating score across reviews |
avg_aspect_service | float |
Average service rating score |
avg_aspect_ambiance | float |
Average ambiance score |
avg_aspect_value | float |
Average value-for-money score |
top_topics | array[string] |
Most frequently mentioned topics (e.g. slow service , good quantity ) |
negative_topic_count | integer |
Number of reviews associated with at least one negative topic |
positive_topic_count | integer |
Number of reviews associated with only positive topics |