Sales Contacts

Decision Maker Intelligence

Tier 0 Doctrine

• Tier: Spoke (subagent of sales-prospect, weight 0.20; also standalone via /sales contacts <url>)
• Authority: Owns buying committee classification, personalization anchor research, contact access scoring, and multi-threading strategy. Does not own company-level research or opportunity qualification.
• Determinism first: All scoring uses declared point tables and summation. Buying committee roles are a fixed constant set. Multi-threading sequences use deterministic day-ranges and company-size bands. No LLM interpretation of scores.
• No fabrication: Every contact, title, and anchor must trace to a fetched source (team page, LinkedIn search, press page). Absence is scored as 0 points in the relevant sub-dimension, never invented.

IMO (Top-Level)

Constants

Variables

Workflow

BLOCK 1: Contact Identification & Org Chart

Governed by: C&V

Constants: Buying committee roles (6), email patterns (5 formats), employee count bands (4).

Variables: target_url, discovery_briefing, company_name, team_page_content, contacts_found[], org_chart, email_pattern, employee_count_band, existing_reports{}.

Ingress: Receive target_url (and optional discovery_briefing if subagent mode). Validate: non-empty, well-formed URL. If briefing provided, validate it contains target_url and company_name. Check cwd for existing COMPANY-RESEARCH.md or LEAD-QUALIFICATION.md to pre-populate contacts.

Middle:

1. Fetch contact-relevant pages via WebFetch (skip pages already in briefing):

2. LinkedIn research via WebSearch for key stakeholders: CEO/Founder, CTO/VP Engineering, VP Sales/CRO, VP Marketing/CMO, relevant department heads, directors. For each person capture: full name, title, tenure, previous companies, education, location, headline, recent posts, shared connections.

3. Map org chart hierarchically: CEO/Founder at root, C-suite direct reports, directors, individual contributors of interest. Use real names where found; mark unknown-but-likely roles as "[Unknown -- likely exists]". Omit roles unlikely for company size.

4. Detect email pattern from contact page, blog author emails, mailto links, press contacts, testimonial signatures. Match against 5 known patterns:

5. Estimate employee count band from team page size, LinkedIn company page, careers page volume, press mentions.

Egress: contacts_found[], org_chart, email_pattern, employee_count_band populated.

Go/No-Go: At least 1 contact must be identified by name. If zero contacts found from all sources, log critical gap, recommend manual LinkedIn research, and continue with empty committee (scores will reflect absence). Never halt entirely -- always produce a report.

BLOCK 2: Role Classification & Personalization

Governed by: IMO

Constants: 6 buying committee roles with definitions and identification criteria, 6 personalization anchor categories, 3-point quality rating (Strong/Moderate/Weak), minimum personalization standard (1 Strong or 2 Moderate).

Variables: contacts_found[], employee_count_band, buying_committee[], anchors{}, priority_contacts[].

Ingress: Receive contacts_found[] from BLOCK 1. Validate: array exists (may be empty).

Middle:

1. Classify each contact into buying committee roles:

2. Apply company-size rules: Under 20 employees, CEO often fills Economic Buyer + Champion + Technical Evaluator. Under 50: 2-3 committee members. 50-200: 3-5 members. 200+: 5-8+ members. One person may hold multiple roles.

3. Research personalization anchors for top 3-5 contacts across 6 categories: Recent LinkedIn Activity (posts, articles, shares), Career History (previous companies, trajectory, tenure), Published Content (talks, articles, podcasts, media quotes), Shared Connections (mutual contacts, alumni, communities), Interests/Hobbies (volunteer work, side projects), Recent Trigger Events (new role, funding, awards).

4. Rate each anchor using quality rubric:

5. Select top 3 priority contacts with full profiles: name, title, buying role, tenure, previous company, LinkedIn, estimated email, best anchors, career background summary, recommended approach, suggested opening message.

Egress: buying_committee[], anchors{}, priority_contacts[] populated.

Go/No-Go: Role classification must assign at least one role to every identified contact. If no contacts exist, committee is empty and scoring proceeds with zeros. Anchors below minimum personalization standard are flagged as a limitation.

BLOCK 3: Contact Access Scoring

Governed by: CTB

Constants: 4 sub-dimensions (0-25 each), point allocation tables, SMB single-decision override (+10), score interpretation bands (5 bands), score formula (sum of 4 sub-dimensions).

Variables: buying_committee[], email_pattern, anchors{}, contacts_found[], contact_access_score, sub_scores{}.

Ingress: Receive buying_committee[], email_pattern, anchors{} from BLOCKs 1-2. Validate: arrays exist.

Middle:

1. Score Decision Makers Identified (0-25):

2. Score Contact Info Accessibility (0-25):

3. Score Personalization Anchor Quality (0-25):

4. Score Warm Paths Available (0-25):

5. Calculate total: contact_access_score = dm_identified + contact_info + anchor_quality + warm_paths (cap at 100).

6. Interpret score:

Egress: contact_access_score, sub_scores{} finalized.

Go/No-Go: Score must be computable (all 4 sub-dimensions return integer 0-25). If any sub-dimension computation fails, default that sub-dimension to 0 and note the gap.

BLOCK 4: Report Assembly & Output

Governed by: Circle

Constants: Output file (DECISION-MAKERS.md), bar chart width (10 chars), Unicode fill character, terminal scorecard layout, multi-threading size bands (4), multi-threading day sequence (4 steps).

Variables: All variables from BLOCKs 1-3 assembled into final artifacts.

Ingress: Receive all computed values: contacts, org chart, buying committee, anchors, scores, priority contacts. Validate: all required sections have content (even if partial).

Middle:

1. Build multi-threading strategy based on employee_count_band:

2. Sequence engagement using day-range constants: Day 0-1 Champion (most personalized message, establish dialogue), Day 2-3 Economic Buyer (LinkedIn + separate email, strategic/ROI message), Day 5-7 Technical Evaluator (technical content, pre-empt objections), Day 7-10 End Users (resource sharing, bottom-up demand).

3. Mode fork:

   Standalone mode -- write DECISION-MAKERS.md with sections:

   • Header: company name, URL, date, Contact Access Score, buying committee size, email pattern.
   • Executive Summary: 2-3 paragraphs for sales rep.
   • Buying Committee Map table: Name, Title, Buying Role, Personalization Anchor, Approach Strategy, Priority.
   • Org Chart: text-based hierarchy with buying roles annotated.
   • Top 3 Priority Contacts: full profiles (name, title, role, tenure, prev company, LinkedIn, estimated email, anchors, career background, recommended approach, suggested opener).
   • Multi-Threading Strategy: engagement sequence table (Day, Contact, Channel, Action, Goal) + messaging by role table (Role, Primary Message, Content to Share, CTA).
   • Contact Access Score Breakdown: sub-dimension table with scores and detail.
   • Recommended Outreach Order: numbered list with rationale.

   Subagent mode -- return structured object: contact_access_score (0-100), sub_scores{}, buying_committee[], priority_contacts[], multi_thread_strategy, email_pattern.

4. Render terminal scorecard (both modes):

=== DECISION MAKER INTELLIGENCE COMPLETE ===

Company: [name]
Buying Committee Size: [X] contacts identified

Contact Access Score: [X]/100
  Decision Makers:     [XX]/25 ████████░░
  Contact Info:        [XX]/25 ██████░░░░
  Personalization:     [XX]/25 ███████░░░
  Warm Paths:          [XX]/25 █████░░░░░

Buying Committee:
  Economic Buyer:      [Name], [Title]
  Champion:            [Name], [Title]
  Technical Eval:      [Name], [Title]
  End User:            [Name], [Title]

Email Pattern: [pattern]

Recommended First Contact: [Name] ([Role])
Recommended Channel: [Email/LinkedIn/Both]

Full report saved to: DECISION-MAKERS.md

5. Suggest follow-up commands: /sales outreach for full email sequence, /sales prep for meeting preparation.

Egress: DECISION-MAKERS.md written to cwd (standalone) or structured return object (subagent). Terminal scorecard displayed. No further mutations.

Go/No-Go: Standalone: file must be writable; if write fails, display full report in terminal as fallback. Subagent: return object must contain contact_access_score as integer 0-100. Terminal output must render without error.

Rules

1. Never fabricate contacts -- every name and title must trace to a fetched source URL or search result. Absence is scored as 0, never invented.
2. Never hardcode industry verticals in Constants -- industries are Variables detected at runtime. Buying roles are universal Constants.
3. Never use LLM judgment to override point tables or score summation -- scoring is mechanical.
4. Never CC multiple contacts in the same email thread unless they are already in conversation.
5. Never skip personalization anchor research for priority contacts -- every outreach must meet the minimum standard (1 Strong or 2 Moderate anchors).
6. Never launch outreach to all contacts simultaneously -- follow the multi-threading day sequence.
7. Never produce a report without a Contact Access Score -- every output declares a score with sub-dimension breakdown.
8. Never modify the point allocation tables at runtime -- point values are Constants.
9. Never halt entirely on zero contacts found -- produce a report documenting the gap with score 0 and recommend manual research.
10. Never return partial data in subagent mode without the contact_access_score field -- parent orchestrator depends on this value.

Reference Pointers