Changelog

All notable changes to ProductPulse are documented in this file.

[Unreleased]

Comprehensive edge-case and error-handling unit tests across analysis engines
Architecture documentation with layer boundaries, data flow, and ADRs
Human-readable metric catalog guide
OutputSerializer for shared CSV/SQLite artifact serialization
Typed pipeline phase result contracts
Deterministic run stamps on tabular outputs
100% coverage gate in local CI
Scenario analysis, cohort summary, funnel summary, and segment funnel output artifacts
Dashboard pages for Scenario Simulator, Cohort Retention, and Funnel Analysis
Data quality dashboard review page
Decision brief summaries with Markdown exports
Owner workload summaries for handoff queues
Quick-view dashboard presets for filtering and sorting
Data freshness indicators and CSV download for all dashboard tables

Updated project dependencies in pyproject.toml
Moved runtime directory creation out of import-time path constants
Standardized Streamlit imports on package import path
Switched dependency source of truth to pyproject.toml only
Grouped generated metric definitions by catalog category
Added cross-links between generated Markdown reports
Wired previously standalone cohort, funnel, and scenario engines into the main pipeline

Resolved strict mypy import path validation errors in dashboard helper libraries by configuring standard source pathways in pyproject.toml
Fixed unannotated empty set type verification errors in activation funnel calculator
Removed redundant type-casting wrapper warnings in funnel conversion calculations
Corrected stale test count metrics across portfolio case study documents

ProductPulse CLI entrypoint (productpulse run|status|dashboard)
GitHub Actions CI workflow (make ci)
Model contract tests and schema validation
Makefile for all development commands
MIT License
Synthetic data engine with seeded reproducibility
Analytics engines: KPI, Churn Risk, Revenue Leakage, Recommendations
Metric Governance Engine with YAML-driven catalog
Business Health Score Engine (7 domain scorers)
Customer 360 composite view
Intervention Planner with priority scoring
Decision Trace Engine for explainability
Data Quality Scorer
SQLite persistence layer (writer + read-only reader)
Streamlit dashboard with 11 navigation pages
Markdown report generation (executive summary, metric definitions, risk register)
Generated artifacts Git policy documentation