Internal · Invite-only

Know what every Miinto metric means — and ask for the numbers.

Canonical definitions, column-level lineage, and threaded answers for every number, table, and column at Miinto. Plus live Power BI through Claude when docs aren't enough. One layer your whole team can trust.

Sign-up is invite-only. Ask the BI team for an invitation link.

metrics.miintolabs.com

what is our true gmv?12 results

AllMetricsVariantsTablesColumnsdbt

gmv

Parent metric

human verified

Gross merchandise value: total order value sold via Miinto, excluding VAT, before any rejections.

GMV EURGMV EUR LYGMV Local+5 variants

GMV After Rejections

GMV minus rejected order value

contested

Marketing.GMV EUR

Variant of GMV · in Marketing dataset

verified

marts.fct_orders.gross_value_eur

dbt column · 4 PBI columns downstream

dbt

Stop debating the numbers.

Verified definitions, column-level lineage, drift alerts, threaded answers, and live Power BI through Claude — one layer the BI team curates and the whole company can rely on.

Canonical definitions

One business definition per metric. No more 'which GMV is the right one?' — there's one source of truth, with verification states and an audit trail.

gmv

Parent metric

human verified

Gross merchandise value: total order value sold via Miinto, excluding VAT, before any rejections. Currency is the dataset's currency unless the variant says otherwise.

GMV EURGMV EUR LYGMV Local+5 variants

Variant tracking

Same calculation, different currency or period. All linked to one parent.

gmvparent

GMV EUR·verified

GMV EUR LY·verified

GMV Local·verified

Marketing.GMV·verified

Finance.GMV·draft

Drift alerts

When two implementations of the same metric disagree, the system raises an alert.

DAX drift detected

Marketing.GMV ≠ Finance.GMV

open

Definition stale

rejection_rate · 90d unverified

stale

Threaded Q&A

Ask anything on any element. Mark-accepted-answer. Searchable forever.

Mads

2d ago

What is the difference between Order Created Date and Transaction Date?

Mythili

accepted

Order Created Date is when the customer placed the order. Transaction Date is when the event happened — order, rejection, or return — so it's the one to use for financial reporting.

Curate from Claude

Curate metrics from any Claude client using the built-in admin CLI and MCP server. Plain English in, audit trail out.

~/miinto

$ miinto-admin metrics verify gmv

→ fetching draft definition…

→ ✓ DAX hash matches across 4 datasets

→ marked gmv as human_verified

Miinto Metrics MCP Server

Live Power BI data through Claude — not just documentation. Ask in plain English; the MCP runs the right query against the verified metric and Claude answers with real numbers.

What was GMV by market in April?

query_powerbi(metric=gmv_eur, group_by=market, period=2026-04)0.4s

Result · 3 rowslive

€1.42M

€0.98M

€0.71M

€3.11M total — DK led at €1.42M. verified metric

Lineage

From Power BI back to the source row.

Every column on every dashboard traced to its dbt model — through staging, marts, and renames. So when a source changes, you know exactly what breaks before you ship the change.

pass-through, rename, and transformation edges classified per hop
Power BI workspaces linked to dbt models — every Miinto workspace covered
Blast-radius queries via recursive SQL — find every downstream column in milliseconds

dbt

raw.orders.value_eur

stg_orders.amount_eur

fct_orders.gross_value_eur

Power BI

Marketing.GMV EURverified

Finance.GMV EURverified

Operations.GMVcontested

pass-throughrenametransformation

How it works

One trusted layer, three ways in: a web app for browsing and discussion, an MCP server for asking Claude in plain English, and an admin CLI for the BI team. dbt and Power BI feed it.

STEP 01

Ingest

Paradime pulls the dbt manifest. Power BI scans every Miinto workspace. Three pipelines, scheduled and event-chained.

STEP 02

Curate

BI team reviews stubs, links variants to parents, verifies definitions. Every change goes through an audit trail.

STEP 03

Use

Search the panel. Ask in Claude. Trace lineage. Discuss in threads. Same data, same answer, every time.