1. Ghost Kitchen Aggregator
  2. Features
  3. Metrics and provenance
Feature · Metrics and provenance

Metrics and provenance

Read order measures alongside source coverage and calculation validity.

OverviewResult Validity
Ghost Kitchen Aggregator · Metrics and provenance

Result Validity

Metric results

New
MetricPeriod startPeriod endDimension keyValueSource row count
MD-618218 Oct24 OctDK-59482016
MD-144505 Oct13 OctDK-298303
MD-392506 Oct13 OctDK-83133032
MD-183212 Oct19 OctDK-3735509
MD-469718 Oct26 OctDK-42658022
6 of 6 records
DK-594Metric results · MR-45677
Calculated
MetricMD-6182
Period start18 Oct
Period end24 Oct
Value820
Source row count16
Example screen with sample data

What you can do

Read order measures alongside source coverage and calculation validity

An analyst compares kitchen partner performance over a reporting period. The analyst checks snapshot coverage, the metric definition and the result status before using the value. A suppressed or invalid result stays visible with its reason instead of being presented as a normal calculation. The work starts with metric grain and aggregation method. Those details let the next person open the same order or partner context instead of relying on a message thread. The team then checks observation period, snapshot coverage and source row count. The important decision is whether a reported value has a reviewed definition and a valid result state.

The records here separate an operating fact from its review. An order can have a status, an observation can have an outcome, and a run can have an execution result. Each answers a different question. When the team sees a stale snapshot or invalid result, it keeps the definition version and result reason beside the source order so a later reviewer can reconstruct what happened.

How the work is organized

  • Identify the source. Start from metric grain and aggregation method and keep the related organization, partner and order visible.
  • Check the decision. Review observation period, snapshot coverage and source row count before a result is used.
  • Keep the trail. Record the definition version and result reason with the same source identity when a question needs investigation.

What to agree during setup

Agree the metric grain, unit, aggregation method and definition version before reading a result. Metrics Analyst should compare observation periods with snapshot coverage and freshness. A result marked Suppressed or Invalid needs its reason and source row count beside the value.

Give the Metrics Analyst a definition with a grain and aggregation method. Walk through an accepted case and a stale snapshot or invalid result. Check that the person responsible for the next step can find the linked source record and its current status. Keep earlier revisions, outcomes and reasons available when the team investigates a change.

Check the handoff on an actual case

An analyst compares kitchen partner performance over a reporting period. The analyst checks snapshot coverage, the metric definition and the result status before using the value. A suppressed or invalid result stays visible with its reason instead of being presented as a normal calculation. Ask Metrics Analyst to open the definition, source observation and snapshot that support the result. Compare the reporting period with coverage dates and read the result status and reason. Do not present a suppressed or invalid value as a normal calculated result.

Rehearse the case with a returned or uncertain outcome as well. The team should keep that result visible, assign an action with a due date where needed, and close the action with a resolution that points back to the order. The order trail should still show what happened before the correction.

What your team sees

  • Result Validity — Separate calculated results from suppressed or invalid ones.
  • Metric Review Inbox — Find draft metric definitions awaiting review of grain and method.

Follow the work through

Follow the worked process and agree the implementation sequence before the first operating cycle.

What it keeps track of

Data model
Metric definitions
Reviewed metric definitions with explicit grain and aggregation method.
Metric observations
Observation-level metric inputs tied to source orders and snapshots.
Metric results
Workflow-calculated metrics with source row counts and validity state.
Source snapshots
Captured source-system coverage and freshness evidence.
Operation runs
Durable idempotent workflow runs and actual results.
Kitchen orders
Parent order revisions routed through kitchen partners and pooled drivers.

Who uses it

Permissions
Metrics Analyst
, Independent Reviewer and Admin.
The first named role owns the daily handoff; the others review the context or investigate exceptions.

Set up Ghost Kitchen Aggregator your way.

Start free, then use Proto to change fields, steps and approvals before your team goes live.

How to get started