Metric Reconciliation
Traces why the same KPI shows two different numbers in two dashboards, down to the exact step where the definitions diverge.
Overview
Metric Reconciliation kicks in when two dashboards disagree on the same KPI, a metric changed unexpectedly after a pipeline update, or a legacy reporting source needs to be merged or deprecated.
What it covers
- Confirms the exact metric definition (numerator, denominator, filters, time zone) since mismatched definitions are the most common cause of a discrepancy.
- Pulls the metric value from both sources for the same period and computes the absolute and percentage gap.
- Documents an accepted tolerance when the gap is within an agreed threshold, and closes the investigation there.
- Traces each source's computation path step by step to find the divergence point (join type, filter order, null handling, date truncation).
- Classifies the root cause and completes a reconciliation report showing both sources, the gap, and the resolution.
Part of Data Analytics Skills
This is one of 31 skills in the Data Analytics Skills library, in the Documentation & Knowledge group.
Installation
- Clone the repo once:
git clone https://github.com/nimrodfisher/data-analytics-skills.git - Copy just this skill:
cp -r data-analytics-skills/02-documentation-knowledge/metric-reconciliation ~/.claude/skills/(or~/.agents/skills/for Codex) - Ask your agent something like "why do these two dashboards show different churn numbers" and it activates on its own.
See the Data Analytics Skills entry for the full library.
Related assets
Frontend Design
Generates distinctive, production-grade UI that avoids generic "AI slop," with bold typography, cohesive color, and precise motion, implemented as real code.
Graphify
Turns a codebase, plus its docs, PDFs, images, and video, into a queryable knowledge graph that a coding agent can traverse instead of grepping raw files, and never guesses silently: every edge is tagged EXTRACTED, INFERRED, or AMBIGUOUS so you know what was found versus inferred.
Caveman
Makes your coding agent answer in a stripped down, high signal style that keeps every technical fact and cuts filler, saving output tokens on chat style questions.