What fraction of orders exceed a revenue threshold?

Prerequisite

This recipe builds on the fixed-grain order_total_revenue metric from Average order value. Complete that recipe first.

When would you use this?

Operations needs to identify premium trucks - those where more than half of orders exceed $50 - to decide which locations to prioritize for higher-end menu items.
A product manager asks “What share of orders are high-value?” and you need a metric that works across any grouping dimension.
You want a threshold that adapts to the data distribution rather than a hard-coded dollar amount.

What this recipe builds

A ratio metric that counts orders above a threshold and divides by total orders. The threshold is evaluated at order grain using the fixed-grain inner metric (order_total_revenue), then the share is computed at user context. The recipe shows two approaches to defining the threshold:

This recipe uses the ratio metrics concept. See the documentation for the full technical reference.

Use case 1 - Fixed dollar threshold

A hard-coded threshold, that answers: “What percentage of orders exceed 50 dollars” Use this when the business has a specific dollar cutoff that defines “high value” - for example, a promotion tier or a pricing benchmark.

SQL
YAML

COUNT(DISTINCT order_header.order_id)
FILTER (WHERE order_detail.order_total_revenue > 50)
/ NULLIF(COUNT(DISTINCT order_header.order_id), 0)

type: metric
entity: order_detail
name: high_value_order_share
display_name: High Value Order Share
description: |-
  Share of orders where total order revenue exceeds $50.
  FILTER applied on top of a fixed-grain metric: the threshold
  is evaluated per order, then the share is computed at user
  context.
datatype: float
sql: |-
  COUNT(DISTINCT order_header.order_id)
  FILTER (WHERE order_detail.order_total_revenue > 50)
  / NULLIF(COUNT(DISTINCT order_header.order_id), 0)

Example - sliced by country

Seoul 2022, top trucks by AOV:

Truck	AOV	High-value order share
260	$60.46	56.1%
269	$58.38	50.4%
267	$50.63	42.9%
257	$44.93	36.6%
266	$44.90	37.7%

Trucks 260 and 269 have more than half their orders above $50, making them candidates for premium menu expansion.

Use case 2 - Percentile-based threshold

A data-driven threshold that adapts to the distribution. Instead of a fixed dollar amount, the cutoff is the 75th percentile of order revenue. The share answers: “What percentage of orders fall in the top quartile?” Use this when there is no natural dollar cutoff, or when you want the definition of “high value” to move with the data over time.

75th percentile of order revenue

The global 75th percentile, computed with group by () so it stays the same regardless of user grouping. Filters still apply - filtering to a single country gives that country’s 75th percentile.

SQL
YAML

PERCENTILE_CONT(0.75) WITHIN GROUP
(ORDER BY order_detail.order_total_revenue) group by ()

type: metric
entity: order_detail
name: p75_order_revenue
display_name: 75th Percentile Order Revenue
description: |-
  The 75th percentile of order total revenue.
  Uses group by () so the value is always the global percentile,
  unaffected by user grouping dimensions.
datatype: float
sql: |-
  PERCENTILE_CONT(0.75)
    WITHIN GROUP (ORDER BY order_detail.order_total_revenue) group by ()

The current global value is $57.00.

Top quartile order share

Orders above the 75th percentile as a share of all orders. The FILTER clause references p75_order_revenue - the percentile metric from the previous step - so the threshold adapts as the data distribution changes.

SQL
YAML

COUNT(DISTINCT order_header.order_id)
FILTER (WHERE order_detail.order_total_revenue
  > order_detail.p75_order_revenue)
/ NULLIF(COUNT(DISTINCT order_header.order_id), 0)

type: metric
entity: order_detail
name: top_quartile_order_share
display_name: Top Quartile Order Share
description: |-
  Share of orders where total order revenue exceeds the 75th
  percentile. The threshold adapts to the data distribution
  rather than using a hard-coded value.
datatype: float
sql: |-
  COUNT(DISTINCT order_header.order_id)
  FILTER (WHERE order_detail.order_total_revenue
    > order_detail.p75_order_revenue)
  / NULLIF(COUNT(DISTINCT order_header.order_id), 0)

Example

The global top quartile share is approximately 24.8% - close to the expected 25% for a 75th-percentile cutoff. The small deviation comes from orders that land exactly on the percentile boundary.

How the two behave differently

Aspect	Fixed threshold ($50)	Percentile (p75)
Cutoff value	Always $50	Currently $57, adapts over time
Share when AOV rises	Increases - more orders cross $50	Stays near 25% by definition
Best for	Business rules, promotion tiers	Distribution analysis, outlier detection

Key design notes

Concern	Guidance
Threshold on a fixed-grain metric	Both use cases reference `order_total_revenue`, which uses a fixed `GROUP BY (order_header.order_id)`. Without fixed grouping, the threshold would apply to line-item prices instead.
`group by ()` on the percentile	The percentile metric uses `group by ()` so it computes the global percentile. Without it, the percentile would shift per slice, defeating the purpose of a stable cutoff.
Ratio structure	Both metrics divide filtered count by total count. `NULLIF` prevents division by zero when a slice has no orders.

Average order value - prerequisite recipe
Filters - reference on FILTER (WHERE ...)
Ratio metrics - using filtered metrics inside ratio expressions
Fixed empty grouping - locking the percentile with group by ()

Quick Start

Semantic Ecosystem

Modeling Reference

Context Layer

Management

Integrations

Security

Release Notes

What fraction of orders exceed a revenue threshold?

Prerequisite

When would you use this?

What this recipe builds

Use case 1 - Fixed dollar threshold

Example - sliced by country

Use case 2 - Percentile-based threshold

Example

How the two behave differently

Key design notes

​Prerequisite

​When would you use this?

​What this recipe builds

​Use case 1 - Fixed dollar threshold

​Example - sliced by country

​Use case 2 - Percentile-based threshold

​Example

​How the two behave differently

​Key design notes

​Related reading

Prerequisite

When would you use this?

What this recipe builds

Use case 1 - Fixed dollar threshold

Example - sliced by country

Use case 2 - Percentile-based threshold

Example

How the two behave differently

Key design notes

Related reading