Measure Killer Analysis

Back

Measure Killer Analysis

Measure & Column Usage Report

Analyze

Overview

Features

Prompts

Help

Key Features

Every model picks up baggage over time — measures left over from old versions, columns nothing points at, dependency chains no one remembers building. Measure Killer captures all of it, but the export lands as a wall of nested JSON, so answering something as simple as "is this measure safe to delete?" turns into an hour of clicking around. This skill reads the export for you, runs 28 checks across usage, model quality, and DAX complexity, and hands back a plain picture of what to keep, what to retire, and what needs a second look — including which measures are shared across reports, so you never delete one that's still live somewhere else — run it with /mk-analysis.

How it works

Before you start — this skill reads a Measure Killer JSON export, so you make that in Measure Killer first: open Measure Killer from Power BI Desktop's External Tools ribbon, choose Single model and report, click Analyze, then click Export and choose JSON (full steps in Setup below). Put that file in inputs/measure-killer/, or just tell Claude where you saved it and it'll move it for you.

Then type /mk-analysis, point it to your Measure Killer export, and the skill analyzes everything:

Parses the full Measure Killer JSON export
Builds flat inventories of all measures and columns
Pre-computes dependency graphs (forward and reverse)
Evaluates 28 rules across 6 categories
Generates supplementary data: table usage heatmap, unused measures inventory, cross-report distribution

What you get

Full report (.md) — all findings with affected objects
Compact report (.md) — summarized for stakeholder review
Structured data (.json) — for automation
Table usage heatmap — per-table breakdown showing used vs unused columns and measures
Unused measures inventory — with folder, hidden status, and expression preview
Cross-report distribution — which measures are used in which reports

All files follow consistent naming: {project}_mk_analysis_{date}_{time}.md

Explore more in this category