metric_returns_rate

beta

metric table

Return rate (returns per shipped order), disposition mix, and processing cycle time by segment and region.

Definition

Returns rolled up from the per-return `mart_returns` to (return_month × client_segment × site_region × disposition), where `return_month` is `toStartOfMonth(return_date)`. `return_count` counts returns; `return_value_sum` is the summed value of returned goods (Nullable(Decimal(38,2)) — sum() of Decimal(14,2) promotes to Decimal(38,2)); `avg_return_cycle_days` is the average processing cycle time (Nullable(Float64) — avg() semantics). `shipped_order_count` is the shipped-order denominator, materialized from `mart_fulfillment_daily` at the COARSER (return_month × client_segment × site_region) grain (NO disposition) and repeated across the disposition rows for that month/segment/region. `return_rate` is a FORMULA measure (sum(return_count) / nullif(sum(shipped_order_count), 0)). Both feeds coalesce `client_segment` / `site_region` identically to `unknown` so no bucket drops. The rate is APPROXIMATE — returns are matched to their shipping calendar month, not their exact shipping cohort.

Watch out for

highSlicing return_rate by disposition
`return_rate` is APPROXIMATE and the shipped denominator carries NO disposition — `shipped_order_count` is repeated across disposition buckets. Do NOT slice `return_rate` by `disposition` (it would double-count the denominator). Returns are matched to their shipping CALENDAR MONTH, not their exact shipping cohort.
highReading return_count as the return rate
`return_count` is a raw count of returns, not a rate. Request the `return_rate` formula measure — it divides by `shipped_order_count` with `nullif` window-correctly. Use `return_count` for volume / disposition mix, `return_rate` for the rate.
mediumSumming avg_return_cycle_days across buckets
`avg_return_cycle_days` is an avg() measure — the metric layer re-runs avg() over the chosen window. Do not roll it up client-side (summing or averaging pre-averaged values is biased).

Questions this answers

“What is the return rate by segment this year?”
“What is the disposition mix of returns by region?”
“What is the average return processing cycle time by segment?”

Related metrics

fulfillment
Supplies the shipped-order count (`order_count`) that is this metric's return-rate denominator — pair to read returns against upstream fulfillment volume for the same segment / region.

8 of 8

Column

Type

Description

Tests · PII

return_month

Date

First day of the return month (toStartOfMonth of return_date). Grain + sort key.

1 test

client_segment

LowCardinality(String)

Owning-client segment — automotive / chemicals / fmcg / fashion / high_tech / life_sciences; `unknown` when unmapped. Dimension + sort key. Coalesced identically to the shipped feed.

1 test

site_region

LowCardinality(String)

Handling-site region — EU / UK / NA / APAC; `unknown` when unmapped. Dimension + sort key. Coalesced identically to the shipped feed.

1 test

disposition

LowCardinality(String)

Return disposition — restock / repair / recycle / scrap; `unknown` when unmapped. Dimension + sort key. NOT carried by the shipped denominator (do not slice return_rate by it).

1 test

return_count

UInt64

Count of returns in the bucket. Numerator of the `return_rate` formula measure.

1 test

shipped_order_count

UInt64

Shipped-order denominator from `mart_fulfillment_daily` at the coarser (return_month × client_segment × site_region) grain — NO disposition, repeated across the disposition rows. Denominator of the `return_rate` formula measure. 0 when no matching shipped bucket exists.

1 test

avg_return_cycle_days

Nullable(Float64)

Average return processing cycle time (days) in the bucket. Exposed as the `avg_return_cycle_days` measure (avg). Nullable(Float64) per §0.3 (avg() returns Nullable(Float64)).

return_value_sum

Nullable(Decimal(38, 2))

Sum of returned-goods value in the bucket. Exposed as the `return_value_sum` measure (sum). Nullable(Decimal(38,2)) per §0.3 (sum() of Decimal(14,2) promotes to Decimal(38,2)).