Available Context & Tools

@_platform-references/org-variables.md @_platform-references/capabilities.md

Lead Qualification

Why Qualification Matters

Lead qualification is the single highest-leverage activity in sales:

Only 13% of leads become opportunities. Every hour on a bad lead is stolen from a good one.
Unqualified pursuit wastes 67% of sales time. Qualification is the filter that reclaims it.
Speed-to-lead matters exponentially. Contacting within 5 minutes = 100x more likely to connect than waiting 30 minutes. But speed only matters on the RIGHT leads.
Disqualification is a superpower. The best orgs say "no" the fastest to bad fits.

Your job is not to find reasons to say "yes." Your job is to find the truth -- quickly, honestly, and with enough rigor that the rep can act with confidence.

Goal

Score an inbound lead against ICP criteria using multi-layer intelligence and provide a clear qualification tier with reasoning and recommended next action. The output must be decisive enough that a rep can act on it immediately without second-guessing.

Required Capabilities

CRM: Fetch lead data, company information, and existing relationship context
Web Search: Enrich leads with company funding, news, tech stack, and hiring signals

5-Layer Intelligence Model

Each layer adds depth. Execute layers in order; later layers build on earlier findings.

Layer 1: Lead & CRM Context

Gather baseline data from CRM and user input.

Check if lead exists in CRM: execute_action("get_contact", { email: lead_email })
Check company status: execute_action("get_company_status", { company_name })
Look for existing deals: execute_action("get_deal", { name: company_name })
Load Organization Context (ICP criteria, products, value propositions) from the block above

Layer 2: Web Enrichment

Search the web and enrichment APIs for signals not in CRM.

Company enrichment: Search for recent funding rounds, acquisitions, product launches, leadership changes, hiring patterns, and tech stack signals. These directly inform Budget Signals and Company Size dimensions.
Contact enrichment: If AI Ark or Apollo capabilities are available, enrich the contact for title verification, seniority level, reporting structure, social profiles, and career history. This informs Role Authority scoring.
Competitive signals: Check if the company uses a competitor's product (job postings mentioning competitor tools, review site profiles, tech stack detection).

Layer 3: Historical Context (RAG)

Search meeting transcripts and activity history for relevant context.

Past interactions with this company: Prior demos, discovery calls, support conversations, or any touchpoint. If found, this dramatically changes qualification (they already know you).
Similar company win/loss patterns: Search for transcripts from companies of similar size, industry, and stage. What objections came up? What messaging resonated?
Persona pattern matching: Search for calls with similar titles/roles. What did that persona care about? What made them buy or not buy?
Flag when RAG returns nothing: Distinguish between "first interaction" (no history exists) vs. "data gap" (history may exist but is not indexed).

Layer 4: Behavioral Signals

Assess intent signals beyond static firmographic data.

Website activity: Pricing page visits, feature page views, comparison page visits, return visit frequency
Content engagement: Case study downloads, webinar attendance, whitepaper downloads, newsletter opens
Email engagement: Open rates, click-throughs, reply rates on any prior outreach
High-intent actions: Demo requests, free trial signups, contact form submissions with detailed notes
Low-intent indicators: Passive form fills, single blog visit, unsubscribed from nurture

Score behavioral signals using the engagement rubric in references/icp-templates.md.

Layer 5: Qualification Synthesis

Combine all layers into a scored qualification using the framework below. Every score must cite which layer(s) provided the evidence.

Choosing the Right Framework

Select the framework based on context. Consult references/scoring-frameworks.md for detailed comparisons, worked examples, conversion data, and the framework selection decision tree.

BANT (Budget, Authority, Need, Timeline)

Best for: Transactional sales, shorter deal cycles, SMB/mid-market. Simple and fast. Limitation: Surface-level. Tells you IF they can buy, not WHY they should buy from you. Use for deal sizes under $25K and high-volume qualification.

MEDDICC

Best for: Enterprise sales, complex deals, long cycles, multi-stakeholder buying committees. Limitation: Requires significant discovery. Use to deepen qualification AFTER initial scoring confirms the lead is worth pursuing.

The 5-Dimension Scoring Model (Default)

Default framework. Designed for rapid first-pass qualification with incomplete data. Weights dimensions by predictive power based on B2B SaaS conversion data.

Scoring Framework

Score each dimension 1-5. Use 0 for "cannot assess due to missing data." Cite evidence from specific layers.

1. Company Size Fit (weight: 25%)

Strongest single predictor. If ICP says 100-500 employees:

5: Squarely in sweet spot (e.g., 250 employees)
4: Close to ICP, minor deviation (e.g., 80 employees)
3: Edge of range (e.g., 50 employees -- workable but not ideal)
2: Outside ICP but plausible (e.g., 30 employees, well-funded and growing)
1: Far outside range
0: Cannot determine from any layer

2. Industry Fit (weight: 25%)

Determines value proposition resonance and proof point availability.

5: Target vertical with proven case studies
4: Adjacent industry with proven use cases
3: Some traction but limited proof
2: Unproven but not disqualifying
1: Industry mismatch or known poor fit
0: Cannot determine

3. Role Seniority & Authority (weight: 20%)

Title + company size calibration. Cross-reference with Layer 2 enrichment data.

5: Clear decision-maker with budget authority
4: Strong influencer who can champion internally
3: Mid-level with budget influence but needs approval
2: IC -- may be evaluating but cannot decide
1: No buying authority or unclear role
0: Cannot determine

4. Budget Signals (weight: 15%)

Usually indirect. Layer 2 web enrichment (funding, competitor spend) is critical here.

5: Clear budget indicators -- recent funding (Series B+), paying a competitor, explicit budget mention
4: Likely has budget -- company size/stage suggest it, using adjacent paid tools
3: Possible but unconfirmed
2: Constrained signals -- early-stage pre-revenue, recent layoffs
1: Likely no budget
0: Cannot assess

5. Timing & Intent (weight: 15%)

Layer 4 behavioral signals are primary data source. Behavioral signals (what they DID) outweigh demographics (who they ARE).

5: Active evaluation -- demo request, pricing page visit, competitive mention, stated urgency
4: Engaged and showing intent -- solution content, product webinar, pricing page
3: Interested but early -- blog subscriber, general content, industry webinar
2: Cold inbound with no clear intent
1: Very early or no timing signal
0: Cannot determine

Source Quality Weighting

Apply multiplier to final score:

Source	Multiplier	Rationale
Customer referral	1.25x	4x conversion rate vs cold inbound
Partner referral	1.15x	Strong signal, less trust transfer
Demo request (direct)	1.15x	Explicit high intent
Content inbound	1.0x	Baseline
Event/conference	1.0x	Mixed signal
Outbound (cold)	0.9x	Requires more nurturing
Purchased list	0.8x	Lowest quality

Existing Relationship Detection

Before scoring, check CRM for existing context. This changes qualification dramatically:

Already a customer: Expansion/cross-sell opportunity. Route to account manager.
Open deal exists: Different stakeholder on same opportunity. Connect with deal owner AE.
Past customer (churned): Check churn reason. If resolved, score boost. If fundamental, penalty.
Known contact, no deal: Check prior engagement history.
Mutual connections: Shared investors, board members, or advisors = warm path.

Qualification Tiers

Hot (>= 4.0): Fast-track. Book meeting within 24 hours.
Warm (3.0 - 3.9): Good potential, needs nurturing. Follow up within 48 hours.
Cold (2.0 - 2.9): Low priority. Add to nurture sequence.
Disqualified (< 2.0): Does not meet minimum criteria. Log specific reason and close.

Enrichment Chaining

When qualification data is insufficient, chain to other skills for deeper enrichment:

Condition	Chain To	What It Adds
Score 2.5-3.5, medium/low confidence	`lead-research`	Company deep-dive: tech stack, org chart, recent news, competitive landscape
Contact data sparse (no title, no LinkedIn)	`sales-enrich`	AI Ark/Apollo enrichment: verified title, seniority, direct phone, social profiles
Borderline ICP fit, need competitive context	`lead-research` then `competitor-intel`	Full competitive positioning for the specific company

When to chain automatically vs. recommend: If confidence is Low and the lead source is high-value (referral, demo request), recommend chaining in the next_action. If confidence is Medium, note it in missing_info with impact estimate. Never auto-chain on Disqualified leads.

Common Qualification Mistakes to Avoid

Title bias: A VP who casually browsed is worth less than a Manager who requested a demo. Intent trumps title.
Big logo bias: A Fortune 500 with no intent is worse than a 200-person company matching every criterion.
Recency bias: Score methodically, not by how recently the lead arrived.
Optimism bias: Missing data = score 0, not 3. Uncertainty is not evidence of fit.
Sunk cost reluctance: If it scores below 2.0 after enrichment, disqualify it.
Source worship: Referrals get a multiplier boost, not an override of a score of 1.5.

Output Contract

Return a SkillResult with:

data.qualification_score: Weighted score (0.0 - 5.0), source multiplier applied
data.qualification_tier: "hot" | "warm" | "cold" | "disqualified"
data.confidence_level: "high" | "medium" | "low" based on data completeness across all 5 layers
data.scoring_breakdown: Array of dimension scores with dimension, score, weight, reasoning, data_source, layer_sources (which layers contributed)
data.source_multiplier: Multiplier applied with justification
data.qualification_summary: 2-3 sentences. Lead with verdict, then evidence.
data.strengths: Positive indicators (why this lead might convert)
data.concerns: Risk factors or gaps
data.missing_info: Data points that would improve accuracy, with score swing estimate
data.next_action: action, priority, rationale, suggested_owner, timeline
data.existing_relationship: CRM relationship context
data.framework_recommendation: For Hot/Warm, which deeper framework to use for discovery
data.enrichment_data: Object with web search and API enrichment findings (funding, news, tech stack, hiring, stakeholder profiles). Empty object if enrichment unavailable.
data.rag_context_used: Array of historical context items used (transcript snippets, activity summaries, pattern matches). Empty array if no RAG results.
data.behavioral_signals: Array of intent signals observed (each with signal type, strength, and recency). Empty array if no behavioral data.

Graceful Degradation

Failure Mode	Behavior	Output Note
No lead data at all	Ask user for minimum: company name + title	Template provided in response
Only name/email	CRM lookup. If not found, low-confidence preliminary score	"1 of 5 dimensions scorable. Recommend enrichment."
CRM API failure	Score from available data, flag gaps	"CRM lookup failed. Existing relationship unknown."
Web search unavailable	Skip Layer 2, proceed with Layers 1, 3-5	"Web enrichment unavailable. Budget/company signals may be incomplete."
RAG returns nothing	Distinguish first interaction vs data gap	"No prior interactions found. Scoring without historical context."
Behavioral data missing	Score Timing & Intent from source + stated intent only	"No behavioral tracking data. Intent score based on source signal only."
Enrichment API failure	Proceed without enrichment, flag in missing_info	"Contact enrichment failed. Title/seniority unverified."
Borderline score (within 0.3 of tier boundary)	Call out explicitly with swing factor	"Score 2.8 is 0.2 below Warm. Key swing factor: [specific]."
Conflicting data	Explain conflict and resolution approach	"Title says VP but company is 5 people. Scoring authority at 3."
Competitor/student/spam	Immediate disqualification, no scoring	"Disqualified: [specific reason]."

Always return something. A low-confidence score with caveats and a "get more data" next action is better than silence.

Quality Checklist

Before returning, verify:

Guidelines

Use ICP criteria from Organization Context to calibrate scoring. See references/icp-templates.md for ICP templates, example ICPs for 5 business types, scoring calibration examples, anti-ICP patterns, and ICP evolution guidance.
If critical data is missing, score that dimension 0. Do NOT assume a midpoint.
Apply source quality multiplier as described above.
Check for existing CRM relationships and flag prominently.
Be decisive with the tier. "This could be warm or cold depending on..." is not helpful.
Suggest enrichment chaining for borderline leads (score 2.5-3.5, medium/low confidence).
Consult references/scoring-frameworks.md for BANT/MEDDICC worked examples, framework selection decision tree, conversion rate data, and score recalibration guidance.