mcpField
S
signal-correlation-workbench

signal-correlation-workbench

github.com/gtmagentsData & databases42863 installsUpdated 30w ago
Repository

About signal-correlation-workbench

A data-analysis skill for connecting qualitative customer feedback with churn, expansion, adoption, and customer-health signals. It guides source inventory, data joins, correlation analysis, signal scoring, and the packaging of findings into actionable insights.

Key features

  1. 01Inventory surveys, telemetry, CRM, and finance data
  2. 02Plan keys, sampling windows, and normalization
  3. 03Analyze correlations and cohort differences
  4. 04Score signals by volume, recency, severity, and revenue
  5. 05Turn findings into plain-language insights and visuals

Use cases

  1. 01Assessing feedback links to churn or adoption
  2. 02Combining support, usage, and survey data
  3. 03Testing hypotheses about customer-health indicators

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/gtmagents/gtm-agents --skill signal-correlation-workbench

After installing, type/signal-correlation-workbenchin Claude Code to confirm it loaded.

Details

From the project README

Signal Correlation Workbench Skill

When to Use

  • Quantifying the impact of qualitative feedback on churn, expansion, and adoption.
  • Connecting support data, product usage, and survey responses into a unified narrative.
  • Testing hypotheses about leading indicators for customer health.

Framework

  1. Data Inventory – list all relevant sources (surveys, NPS, CSAT, telemetry, CRM, finance).
  2. Join Strategy – map IDs/keys, sampling windows, and normalization rules.
  3. Correlation Analysis – evaluate relationships (Pearson/Spearman), cohort comparisons, regression snippets.
  4. Signal Strength Scoring – combine volume, recency, severity, and revenue exposure.
  5. Insight Packaging – translate stats into plain language, visuals, and actionable levers.

Templates

  • SQL/notebook snippets for merging VoC tags with product/CRM tables.
  • Dashboard layout showing signal volume vs impact.
  • Experiment tracker linking hypotheses to validated outcomes.

Tips

  • Watch for survivor bias; include lost customers when possible.
  • Flag data quality caveats prominently to maintain trust.
  • Pair with synthesize-voc-insights to auto-embed correlations into narratives.

Comments

Markdown supported