How do I trigger the metric definition process?

You can invoke the skill by saying 'define a metric', 'specify metric', or by using the `/define-metric` command within Claude Code.

Does this skill help with SQL implementation?

Yes, the template includes a specific field for a 'Reference query,' which serves as the canonical SQL implementation to ensure analysts always calculate the metric the same way.

Can it help diagnose why a metric is changing?

Yes, it includes a 'Driver Decomposition' feature that breaks down complex metrics into their core components (e.g., Revenue = Users x Orders x AOV) to help identify the root cause of shifts.

Where are the metrics stored after definition?

The skill automatically registers the metric in your workspace under the 'knowledge/datasets/' directory, creating YAML files that follow a standardized schema.

Metric Spec Definition

Name: Metric Spec Definition
Author: ai-analyst-lab

byai-analyst-lab

•

Analíticas y Monitorización

Standardizes data metric definitions to ensure clarity in calculations, segmentation, and business interpretation.

The define-metric skill provides a rigorous framework for defining data metrics, eliminating ambiguity between engineering and business stakeholders. By enforcing a structured template that includes plain-English definitions, exact SQL formulas, threshold alerts, and driver decomposition, it ensures that every metric—from Conversion Rate to MRR—is documented with precision. It further streamlines data governance by automatically registering these definitions into a project's metric dictionary, facilitating more reliable data analysis and team-wide alignment.

Características Principales

01Driver decomposition for advanced root-cause analysis

02Automatic registration in local knowledge-base metric dictionaries

03Multi-dimensional segmentation mapping (Dimensions, Values, Rationale)

04Standardized metric templates including formulas and components

05Automated threshold configuration for health monitoring and alerts

0619 GitHub stars

Casos de Uso

01Setting up data observability guardrails based on historical performance thresholds

02Onboarding new data analysts with detailed documentation on primary data tables and latency

03Creating a canonical 'Source of Truth' for business KPIs to prevent calculation discrepancies

Características Principales

01Driver decomposition for advanced root-cause analysis

02Automatic registration in local knowledge-base metric dictionaries

03Multi-dimensional segmentation mapping (Dimensions, Values, Rationale)

04Standardized metric templates including formulas and components

05Automated threshold configuration for health monitoring and alerts

0619 GitHub stars

Casos de Uso

01Setting up data observability guardrails based on historical performance thresholds

02Onboarding new data analysts with detailed documentation on primary data tables and latency

03Creating a canonical 'Source of Truth' for business KPIs to prevent calculation discrepancies