commercial-forecaster

Generates three-tier SaaS bookings forecasts, cohort NRR projections, and funnel confidence scores.

25.3k|3.6k|Updated Oct 19, 2025
One-click install
npx skills add https://github.com/alirezarezvani/claude-skills --skill commercial-forecaster
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: commercial-forecaster
Source: https://github.com/alirezarezvani/claude-skills/tree/main/commercial/skills/commercial-forecaster
Command: npx skills add https://github.com/alirezarezvani/claude-skills --skill commercial-forecaster

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

Commercial leaders are pressured to present a single forecast number to the board, but single-number forecasts hide assumptions, mask leaky cohorts, and collapse under CFO scrutiny. This Skill produces a defensible three-tier forecast (commit / best-case / pipe-only) with a mandatory assumption block, plus cohort-level NRR/GRR decomposition and per-stage funnel confidence scoring.

Core Features & Use Cases

  • Three-Tier Bookings Forecast: Computes commit, best-case, and pipe-only numbers from pipeline JSON using blended last-4Q/last-12Q conversion rates, time-to-close probability, and stalled-opportunity downweighting.
  • Cohort ARR Projection: Projects per-cohort NRR and GRR over a quarterly horizon and flags leaky cohorts whose NRR falls 5+ points below the trailing-cohort average.
  • Funnel Confidence Scoring: Scores each funnel stage by coefficient of variation across 12 quarters and assigns confidence bands (HIGH/MEDIUM/LOW/VERY LOW) with treatment recommendations.
  • Use Case: A CRO preparing a QBR fills the intake template with CRM pipeline data, historical conversion rates, and cohort retention, then runs the three scripts to assemble a board slide showing the three numbers, the assumption block, the cohort heatmap, and per-stage confidence.

Quick Start

Fill the forecast intake template with your pipeline, conversion, and cohort data, then ask the assistant to run the bookings forecaster, cohort projector, and funnel scorer to build your quarterly board forecast.

Frequently Asked Questions about commercial-forecaster

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a SaaS bookings forecast for a board meeting?

Export your pipeline from your CRM into the intake JSON with stage, amount, close date, and age per opportunity, then run bookings_forecaster.py. It outputs commit, best-case, and pipe-only numbers with a mandatory assumption block naming the conversion rate, data window, and weighting.

How to detect leaky cohorts in NRR before they hit the consolidated number?

Run cohort_arr_projector.py with per-cohort starting ARR and quarterly retention and expansion percentages. It flags any cohort whose mean NRR falls 5 or more percentage points below the trailing-cohort average, surfacing leaks 2-3 quarters early.

What conversion window should I use for sales forecasting?

The skill blends 70% last-4-quarter conversion with 30% last-12-quarter conversion. Last-12Q alone hides recent regime change, while last-4Q alone overfits a single bad quarter; the blend is disclosed in every output.

Does the commercial forecaster require external Python packages?

No, all three scripts are stdlib-only Python with no third-party dependencies. Each supports --help and --sample flags, so you can test with built-in sample data before supplying your own intake JSON.

When should I not use this forecasting skill?

Do not use it for backward-looking financial close or GAAP reporting, multi-year strategic financial planning, pricing decisions, or per-deal discount approval. It is strictly forward-looking pipeline and cohort math for the current forecast horizon.

Why is my commit forecast flagged as unsupported?

The tool warns when pipeline coverage falls below 3.0x, meaning total pipeline dollars divided by the commit number is under the SaaS-industry floor. It also flags sandbagging when best-case is under 50% of pipe-only and hockey-sticking above 80%.