commercial-discovery
SKILL.md
Commercial Discovery (B2B Consulting Sales)
Purpose
- Enable thorough, structured discovery for B2B consulting sales.
- Unlike SaaS discovery (which demos a product), consulting discovery must deeply understand the client's current state, desired future state, organizational dynamics, and constraints to later design a custom solution.
- Prepare the seller and capture structured notes.
Key Differentiation from SaaS Discovery
- No product to demo — discovery IS the product sample.
- Must assess organizational readiness, not just feature fit.
- Need to understand current tech stack, team capabilities, and culture.
- Must map multiple stakeholders (not just a single buyer).
- Consulting discovery often spans 2-3 meetings, not one.
Scope
-
This skill WILL:
- Generate pre-meeting research briefs with tailored SPIN questions
- Map buying committees and stakeholder dynamics
- Produce structured post-meeting discovery notes
- Run mini tech-maturity assessments during discovery
- Update pipeline state with discovery insights
-
This skill WILL NOT:
- Propose solutions (defer to solution-design phase)
- Generate proposals or SOWs
- Conduct qualification scoring (see commercial-qualification)
Inputs
- prospect-profile.md — from commercial-prospecting
- commercial-state.md — pipeline context
- user_input — meeting details, known contacts, specific areas to explore
Outputs (contract)
Output 1: Pre-Meeting Brief (discovery-prep.md)
- Company research summary — key facts, recent news, strategic context
- Known pain points and hypotheses — from prospecting or prior interactions
- Stakeholder map — known contacts, roles, likely agenda
- SPIN question guide — 15-20 questions tailored to this prospect, organized by S/P/I/N (see
references/discovery-frameworks.md) - Meeting agenda suggestion — 45-60 min structure
- Red flags to watch for — signals this opportunity may not be real
- Success criteria for the meeting — what "good" looks like
Output 2: Post-Meeting Discovery Notes (discovery-notes.md)
- Meeting metadata — date, attendees, duration
- Current State summary — tech stack, processes, team, pain points
- Desired Future State — what success looks like for them
- Gap Analysis — current → desired, organized by Software / Data / AI
- Buying Committee Map — Champion, Economic Buyer, Technical Buyer, Coach, Blocker — with names
- Budget signals — explicit mentions, inferred range
- Timeline signals — urgency drivers, deadlines, fiscal year
- Competition signals — other vendors mentioned, internal alternatives
- Next steps agreed
- Open questions requiring follow-up
Output 3: Updated commercial-state.md
Update the opportunity with discovery insights: stage, champion, key pain points, next action.
SPIN Framework Adapted for Tech Consulting
- Situation: Current tech landscape, team structure, processes, recent initiatives
- Problem: Pain points, inefficiencies, failed past initiatives, technical debt
- Implication: Business impact of not solving (revenue loss, competitive risk, team attrition, compliance risk)
- Need-payoff: Value of solving (ROI, speed, capability unlock, market advantage)
For the full SPIN question bank organized by service line (Software, Data, AI) with 10 questions per category, see references/discovery-frameworks.md.
Mini Tech Maturity Assessment (during discovery)
- Run a quick 5-question assessment per axis (Software / Data / AI) to validate or update prospecting scores.
- Compare self-reported maturity vs. observed indicators.
- Full questionnaire available in
references/discovery-frameworks.md.
Guardrails (must follow)
- Discovery is about listening, not pitching — question-to-statement ratio should be 3:1 minimum.
- Never propose a solution during discovery — note the urge, defer to solution-design phase.
- Always map at least Champion + Economic Buyer — if you cannot identify both, flag as risk.
- Capture exact quotes when possible — client's own words are gold for proposals.
- Never assume budget — probe with indirect questions.
- If discovery reveals the prospect is not a fit, say so honestly rather than forcing it.
- Flag when a single discovery meeting is insufficient and recommend follow-up.
Example
Context: Logistics company (Acme Logistics) exploring data platform modernization. Legacy SQL Server data warehouse, 15-person IT team, $40M revenue.
Sample SPIN Questions:
Situation:
- "Walk me through how data currently flows from your TMS and WMS into the SQL Server warehouse."
- "How many people on the team write queries or reports against the warehouse today?"
Problem:
- "What happens when leadership asks for a report that crosses multiple source systems?"
- "How long does it take to onboard a new data source into the warehouse?"
Implication:
- "When route optimization decisions are delayed because data isn't ready, what's the cost per day in fuel and driver hours?"
- "If the warehouse goes down during peak shipping season, what's the operational impact?"
Need-payoff:
- "If your operations team had real-time visibility into shipment status across all carriers, how would that change your customer SLA performance?"
- "What would it mean for the business if you could add a new data source in days instead of months?"
Weekly Installs
5
Repository
piperubio/ai-agentsGitHub Stars
1
First Seen
Feb 23, 2026
Security Audits
Installed on
amp5
github-copilot5
codex5
kimi-cli5
gemini-cli5
cursor5