maiven gateway · online··
manifest · 1d ago
Sign in Back to catalog
Definition
Per-product review summary joined to the catalog — total / verified / low / high rating counts, average rating, helpful votes, and a coarse `quality_tier` flag (`needs_attention` / `well_reviewed` / `lightly_reviewed`). Includes un-reviewed products so BI can spot catalog gaps. Snapshot, not a trend — for the moving picture use `mart_product_reviews_daily`.
Watch out for
- highReading `avg_rating` for un-reviewed products
`avg_rating` is null when `total_reviews = 0` — it is NOT zero. Filter `total_reviews > 0` before averaging or you will treat a missing rating as a bad one.
- mediumUsing this snapshot for a ratings trend
`mart_product_quality` is a point-in-time per-product summary. For "are ratings drifting over time" use the `product_reviews` metric (`mart_product_reviews_daily`).
Questions this answers
- “Which products need attention (low ratings, enough reviews)?”
- “How many catalog products have zero reviews?”
15 of 15
Column
Type
Description
Tests · PII
product_id
UInt64
Primary key — UInt64. Matches `stg_postgres__products.product_id`.
3 tests
sku
String
Stock-keeping unit code — unique business identifier carried from the catalog.
2 tests
product_name
String
Human-readable product name from the catalog.
category
Nullable(String)
Product category from the catalog. May be null on the source; left as-is.
1 test
unit_price
Decimal(10, 2)
Catalog list price. Decimal(10,2).
is_active
Bool
Whether the product is currently sellable in the catalog.
total_reviews
UInt64
Count of reviews matched to this product (zero for un-reviewed products).
2 tests
verified_reviews
UInt64
Of `total_reviews`, how many came from a verified-purchase reviewer.
2 tests
low_rating_count
UInt64
Of `total_reviews`, how many gave a rating ≤ 2.
2 tests
high_rating_count
UInt64
Of `total_reviews`, how many gave a rating ≥ 4.
2 tests
avg_rating
Nullable(Float64)
Float 1.0..5.0 — null when total_reviews=0.
1 test
total_helpful_votes
UInt64
Sum of `helpful_votes` across the product's reviews. Zero when un-reviewed.
first_review_at
Nullable(DateTime64(3))
Timestamp of the product's earliest review. Null when un-reviewed.
last_review_at
Nullable(DateTime64(3))
Timestamp of the product's most recent review. Null when un-reviewed.
quality_tier
String
Bucketed quality flag — `needs_attention` when ≥5 reviews and avg rating ≤ 2.5; `well_reviewed` when ≥10 reviews and avg rating ≥ 4.0; `lightly_reviewed` otherwise (includes un-reviewed products).
2 tests