Lead Scoring

Lead Scoring

When to Use

Activate when a founder needs to evaluate inbound prospects against ICP criteria, build a systematic qualification workflow, score and route leads, establish MQL/SQL definitions, or design pipeline stages. Also use when the user says "which leads should I focus on," "how do I qualify inbound leads," "define my ICP," "set up lead scoring," or "how do I route leads to the right person."

Context Required

From startup-context or the user:

• ICP definition — Who is the ideal customer (company size, industry, stage, geography, use case)
• Lead sources — Where inbound leads come from (website, events, content, referrals)
• CRM and tooling — Current stack for managing leads and deals
• Current customers — Who are the best existing customers and why
• Pipeline data — Existing deals, active customers, prior contacts
• Sales capacity — Who handles leads and what is their bandwidth

Work with whatever the user provides. If they have a clear problem area, start there. Do not block on missing inputs.

Workflow

1. Load ICP and configuration — Read startup-context if available. Establish the qualification criteria across company attributes, person attributes, and use case fit.
2. Parse the lead data — Accept leads in any format (CSV, list, CRM export, single name). Identify data gaps and flag what needs enrichment.
3. Check pipeline overlap — Before scoring, check for existing customers (route to upsell), active deals (flag for sales coordination), and prior contacts (note history). Pipeline overlaps are routing flags, not disqualifiers.
4. Score company fit — Evaluate against company size, industry, stage, geography, and use case alignment. Weight each dimension based on what predicts closed-won deals.
5. Score person fit — Evaluate title, seniority, department, and decision-making authority. A perfect company with the wrong contact still needs routing, not rejection.
6. Score use case alignment — Connect the lead's inferred intent to specific product capabilities. Inbound signals (demo requests, pricing page visits) tip borderline cases toward qualification.
7. Generate composite score and verdict — Produce a 0-100 composite score and assign a routing recommendation.
8. Export structured output — Deliver results in a table or CSV with all qualification data, scores, and routing.

Output Format

Deliver these documents:

1. Scored lead report — Each lead with composite score (0-100), sub-scores by dimension, verdict category, and routing recommendation
2. ICP definition — Firmographic and demographic criteria with priority tiers
3. Scoring model — Complete point-value table for company, person, and use case dimensions with threshold definitions
4. Pipeline routing rules — How each verdict category gets handled

Frameworks & Best Practices

Verdict Categories

Assign every lead to one of these routing buckets based on composite score:

| Verdict | Score | Action | |---------|-------|--------| | Qualified — Hot | 85-100 | Immediate sales outreach. High urgency, strong fit. | | Qualified — Warm | 75-84 | Active pursuit within 24 hours. Good fit, moderate urgency. | | Borderline | 50-74 | Requires human review. Qualified with caveats — flag specific concerns. | | Near Miss | 30-49 | Nurture sequence or referral opportunity. Not ready for sales. | | Disqualified | 0-29 | Does not fit ICP. Includes competitor employees. Polite decline. |

Handling Unknown Data

Score unknown dimensions at 30 points (out of 100 for that dimension). This acknowledges data absence without automatically rejecting leads. A lead missing company size data is not the same as a lead with the wrong company size. Flag unknowns for enrichment rather than penalizing them.

Inbound Intent Premium

Prospects who initiate contact demonstrate genuine interest. For borderline cases (scores 50-74), inbound signals should tip the scoring decision toward qualification. A borderline lead who requested a demo is a better prospect than a slightly-above-threshold lead who has never engaged.

Pipeline Overlap Routing

Before scoring, check for overlaps and route accordingly:

• Existing customer — Route to account management for upsell/expansion conversation
• Active deal in pipeline — Flag for the assigned sales rep to coordinate, do not create a duplicate
• Prior contact with no deal — Note history and score normally, but include context for the sales rep
• Competitor employee — Auto-disqualify and log for competitive intelligence

Multi-Dimensional Scoring

Company evaluation — Score against: company size, industry vertical, company stage/funding, geography, and use case fit. Weight dimensions based on which most predict closed-won deals in your data.

Person assessment — Score against: job title, seniority level, department alignment, and decision-making authority. A Director of Engineering at a perfect-fit company scores higher than a junior developer at the same company.

Use case alignment — Map the lead's stated or inferred needs to specific product capabilities. Strong alignment on the core use case matters more than broad but shallow fit.

Dual-Threshold MQL Definition

An MQL requires BOTH fit and engagement. Neither alone is sufficient.

• Minimum fit score: 30 points (must have basic ICP match)
• Minimum engagement score: 20 points (must show some intent)
• Combined minimum: 60 points

A perfect-fit company that never engages is not an MQL. A student downloading every whitepaper is not an MQL. The dual-threshold prevents both failure modes.

Maintaining and Iterating

• Recalibrate quarterly. Pull closed-won data and check if the model correctly predicted winners.
• Watch for score inflation. If 80% of leads become MQLs, the threshold is too low.
• Track MQL-to-SQL acceptance rate. If sales rejects more than 30% of MQLs, adjust the model.
• Start simple. Score the first 50-100 leads by hand before automating.
• Speed-to-lead is critical. Contact within 5 minutes is 21x more likely to qualify.

Related Skills

• cold-outreach — Use the ICP and scoring to prioritize who to reach out to first
• sales-script — Use pipeline stage definitions to prepare the right script for each stage

Examples

Example prompt: "We get 200 inbound leads a month from our website and events. Most go nowhere. Help me build a system to score and route them."

Good output excerpt: > ### Lead Qualification Report (Sample) > | Lead | Company Score | Person Score | Use Case Score | Composite | Verdict | > |------|-------------|-------------|---------------|-----------|---------| > | Jane Smith, VP Eng @ Acme (200 emp, SaaS) | 88 | 85 | 90 | 88 | Qualified — Hot | > | Bob Lee, Developer @ TinyCo (15 emp, Agency) | 35 | 40 | 50 | 40 | Near Miss | > | Unknown Title @ MegaCorp (10K emp, Finance) | 60 | 30 (unknown) | 45 | 47 | Near Miss — Enrich | > > Routing: Jane gets immediate sales outreach (AE assigned within 1 hour). Bob enters nurture sequence. MegaCorp lead flagged for enrichment — title and use case data needed before routing.

Example prompt: "A lead from a current customer's company just filled out our demo form. What do I do?"

Good output approach: Flag the pipeline overlap — check if this is a new department/team or the same buyer. If same account, route to the existing account manager for upsell coordination. If new department, score normally but include account context. Never create a duplicate deal.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: shawnpang
• Source: shawnpang/startup-founder-skills
• License: MIT

Install and usage instructions live in the source repository linked above.