--- name: client-dossier version: 1.0.0 description: > Triggers on "build dossier on", "client profile", "prospect research", "research company X", "build profile on X person", "dossier for sales call", "quién es X empresa". Produces a branded HTML dossier: company DNA, key contacts, pain points, entry strategy. argument-hint: "company-name or person-name" model: opus context: fork allowed-tools: - Read - Write - Edit - Glob - Grep - WebSearch --- # Client Dossier — Intelligence Brief for High-Stakes Meetings > TL;DR: Transform public signals about a prospect into a battle-ready dossier — company DNA, > key contacts, pain hypotheses, and a personalized approach strategy. Output: branded HTML > dossier (JM Labs dark style) + optional markdown summary. **Principio Rector:** Prepare like a detective, enter like a consultant. Every fact verified, every hypothesis labeled. A dossier is not an encyclopedia — it is a decision support tool. --- ## When to Activate **Activate when:** - User says "build dossier on [company/person]", "research [company X]", "client profile for [name]" - User says "quién es [empresa]", "investiga a [empresa]", "perfil de prospecto para [nombre]" - User says "dossier for sales call", "prep for meeting with [company]", "who is [person] at [company]" - A sales or BD context is active and a company or executive name is the primary input [INFERRED] - User provides a company name and asks for entry strategy, pain points, or stakeholder map [INFERRED] - User has a discovery call in <48h and needs preparation material [INFERRED] **Do NOT activate when:** - User is asking for general market research without a named target (use market-intelligence instead) - The request is about internal team members or employees (use stakeholder-mapping instead) - The request is about a private individual in a non-business context (privacy concern — decline) - User wants a competitive landscape across multiple players (use competitive-intelligence instead) - Input is too vague: "research my industry" — no named target present, clarify first - User needs legal/financial due diligence for M&A (requires specialized legal tools, not this skill) --- ## S1: Dossier Types Three dossier modes, each with a distinct research agenda. Infer the mode from phrasing. If ambiguous, ask: "Is this a company dossier, an executive profile, or a partnership evaluation?" **Company Dossier** [EXPLICIT] Full organizational profile. Used before a first meeting, RFP response, or partnership pitch. Covers founding story, ownership, revenue signals, headcount trajectory, tech stack, recent strategic moves, and cultural tone. Most common mode. Output: all 6 sections (S2–S5 + HTML). **Executive / Public Figure Dossier** [EXPLICIT] Profile of a named decision-maker. Used to personalize outreach, prepare for a call, or understand a gatekeeper. Covers career arc, public positions, speaking topics, known priorities, management style signals, and mutual connections. Strict scope: public professional information only. No personal life, no private data inference. Do not include home addresses, personal social media, family data, or non-professional history. **Partnership Target Dossier** [INFERRED] Hybrid of company + lead contact. Used when evaluating a channel partner, reseller, or co-builder. Emphasizes business model alignment, revenue complementarity, integration potential, and historical partnership behavior: who they partner with, on what terms, and how long those partnerships last. **Minimum viable input per mode:** | Mode | Required | Useful extras | |---|---|---| | Company | Company name | Website, LinkedIn URL, industry | | Executive | Full name + company | LinkedIn URL, role title | | Partnership | Company name + relationship type | Customer segment, known tech stack | **Dossier depth options:** | Mode | Time | Scope | Use case | |---|---|---|---| | Flash (30 min) | Fast | Company DNA + 1 contact | Inbound call in <1h | | Standard (2h) | Medium | Full S2–S5, 2–3 contacts | Tomorrow's discovery call | | Deep (5h+) | Slow | Full + org chart + competitive context | Strategic account, large deal | --- ## S2: Company DNA Framework Systematic extraction of organizational identity signals. Work through each dimension. Tag every claim with `[EXPLICIT]`, `[INFERRED]`, or `[OPEN]`. [EXPLICIT] **Founding & Ownership** - Year founded, founding team background (operators vs. investors vs. technologists) [EXPLICIT] - Current ownership: bootstrapped, VC-backed (series + lead investor), PE-owned, public [EXPLICIT] - Acquisition history: have they been acquired or acquired others? Signals strategic ambition [EXPLICIT] - Ownership concentration: founder-led vs. institutional vs. family-controlled [INFERRED] **Revenue & Growth Signals** - Public revenue if available; otherwise triangulate from funding rounds + headcount trend [INFERRED] - Growth stage: pre-PMF, scaling, mature/optimizing, declining/pivoting [INFERRED] - Geographic footprint: single market, regional, global — matters for solution scope [EXPLICIT] - Hiring velocity: net new roles in last 90 days signals investment in capability [INFERRED] **Headcount & Team Composition** - Total headcount from LinkedIn (snapshot, not exact); engineering % vs. sales % vs. ops % [INFERRED] - Departures / executive churn: LinkedIn tenure data, Glassdoor reviews as qualitative signal [INFERRED] - Key department gaps: 5 open data engineer roles → they have an infrastructure problem [INFERRED] - Note: headcount from LinkedIn can be ±15-20% of actual; label as `[est.]` [EXPLICIT] **Tech Stack Signals** - Job posting language: "must know Kubernetes", "experience with Salesforce", "dbt + Snowflake" - BuiltWith / Wappalyzer (if web-facing product) for frontend/marketing stack [INFERRED] - GitHub org activity: open-source repos, language distribution, commit frequency [EXPLICIT] - Integration partners mentioned on their website or in press releases [EXPLICIT] **Recent News & Strategic Moves (last 90 days)** - Funding announcements, acquisitions, product launches [EXPLICIT] - Leadership changes (new CTO/CPO = new agenda, new vendor evaluation cycle) [INFERRED] - Customer wins / case studies published = proof of strategic direction [EXPLICIT] - Conference appearances and speaking slots = what narrative they are pushing publicly [INFERRED] - Regulatory or legal events (SEC filings if public, lawsuit mentions) [EXPLICIT] **Cultural & Operational Tone** - Mission language on careers page: "move fast", "customer obsessed", "enterprise-grade" [INFERRED] - Glassdoor themes: execution culture vs. churn risk; note review recency [INFERRED] - Social media tone: thought leadership vs. pure broadcast vs. community-engaged [INFERRED] **Financial Signals Table:** | Signal | What It Indicates | |---|---| | Recent funding round | Available budget, growth phase, investor expectations | | High job posting volume | Investment areas, team priorities, active pain points | | Contract announcements | Revenue tiers, deal size norms | | Downsizing / layoffs | Budget pressure, cost-saving motive, churn risk | | Awards / rankings | Brand health, employer strength, strategic priorities | | Re-posted same role 3× | Retention problem or unrealistic hiring bar | > [OPEN] Revenue figures for private companies are estimates. Flag all inferred revenue > with `[est. — not verified]` in every output. Never present as confirmed facts. --- ## S3: Key Contacts Map Build a stakeholder map of the target organization focused on the deal or engagement type. Produce a Contact Card (see template below) for each named person. [EXPLICIT] **Decision Maker Identification by Deal Type** | Deal Type | Decision Maker | Influencer | Gatekeeper | |---|---|---|---| | Technology / SaaS | CTO, VP Engineering | Engineering Manager | EA, Procurement | | Consulting / services | CEO, COO | CFO | EA, Legal | | Training / enablement | CHRO, VP L&D | Department Head | Training Coordinator | | Data / analytics | CDO, VP Data | Data Lead | Procurement | | Marketing tech | CMO, VP Marketing | Demand Gen Lead | Marketing Ops | | Digital transformation | CIO / CDO + C-suite sponsor | COO | PMO, Legal | **Contact Research Protocol** (execute in this order): 1. LinkedIn search: "[Company] [Title]" — note tenure, previous companies, education [EXPLICIT] 2. Company website team/about page — often lists leadership with bios [EXPLICIT] 3. Search "[Name] [Company] interview" or "[Name] [Company] podcast" for public positions [EXPLICIT] 4. Twitter/X profile for thought leadership signals and engagement patterns [EXPLICIT] 5. Note mutual connections — warm intro is always the highest-value outreach path [INFERRED] **Email Pattern Inference** (validate before use — never assume confirmed): - `firstname@company.com` — common at startups and SMB - `firstname.lastname@company.com` — most common at mid-market and enterprise - `flastname@company.com` — common in finance and professional services - `f.lastname@company.com` — frequent in European companies Use Hunter.io pattern matching (if user has access) or validate via email verification tool. Label all email patterns as `[INFERRED — not verified]` in output. [EXPLICIT] **Contact Card Template (produce one per person):** ``` Name: [Full Name] [EXPLICIT] Title: [Current Title] [EXPLICIT] Tenure: [X years at company] [EXPLICIT] Prior companies: [2–3 most relevant past employers] [EXPLICIT] Background: [2-sentence career arc] [EXPLICIT] Public positions: [Key theme they discuss publicly] [EXPLICIT] Best channel: [LinkedIn / email / warm intro via X][INFERRED] Likely priority: [What they care about professionally][INFERRED] Potential concern: [What might make them say no] [INFERRED] Email pattern: [pattern@company.com — confidence: H/M/L] [INFERRED] Mutual connection: [Name + relationship if applicable] [EXPLICIT / OPEN] ``` **Org Chart Reconstruction** [INFERRED] - Map reporting lines from LinkedIn titles and tenure signals - Identify recently promoted people (likely championing new initiatives) - Flag who joined from a competitor (may bring vendor preferences with them) - Note who is actively hiring under them (signals budget ownership) > [EXPLICIT] Only use publicly available professional information. Do not infer personal > details, home addresses, or non-professional social media content. --- ## S4: Pain Point Hypothesis Construct a hypothesis of what is hurting the target organization right now. This drives message-market fit for the outreach or pitch. Produce 3–5 hypotheses. [EXPLICIT] **Signal-to-Pain Mapping:** | Signal Observed | Derived Pain Hypothesis | |---|---| | 15+ open engineering roles | Scaling pains, team capacity issues, delivery bottleneck | | 3 CTOs in 24 months | Tech strategy instability, culture/direction tension | | No data/analytics roles | Low data maturity, spreadsheet-heavy operations | | Series B raised 12 months ago | Board pressure for growth metrics and unit economics | | Recent compliance-related press | Regulatory risk, need for audit/governance tooling | | Legacy tech stack in job postings | Migration need, tech debt, developer experience friction | | Rapid expansion to new markets | Localization needs, operational complexity, compliance gaps | | Re-posting the same role 3× | Retention problem or misaligned expectations | | Competitor launched a feature they lack | Product gap, pressure to close the gap fast | | Customer forum / G2 complaints visible | Support overload, churn risk, UX debt | **Pain Point Hypothesis Format (produce per hypothesis):** ``` Hypothesis: [Company X] likely struggles with [SPECIFIC PAIN] [INFERRED] Evidence: [Signal 1: source] + [Signal 2: source] Urgency: [Why this pain is acute NOW vs. chronic background noise] [INFERRED] Our hook: [How our offering maps to this specific pain] [INFERRED] Confidence: High / Medium / Low Validation Q: [The one question to ask on the call to confirm or refute] ``` **Pain Categories Checklist:** - [ ] Revenue growth blocked (can't acquire, expand, or retain customers) - [ ] Operational efficiency (manual work, slow processes, high headcount for low output) - [ ] Technical debt (legacy stack slowing product velocity) - [ ] Talent (can't hire, retain, or upskill fast enough) - [ ] Compliance / risk (regulatory pressure, security gaps, audit failures) - [ ] Data visibility (can't measure what matters, decisions made on gut) - [ ] Customer experience (churn, NPS decline, support volume growing) > [EXPLICIT] Pain hypotheses are hypotheses. They must be validated in the discovery call. > Do not present them as confirmed facts in the dossier or in outreach messages. --- ## S5: Approach Strategy Synthesize dossier findings into a concrete outreach and engagement plan. [INFERRED] **Best Entry Angle Selection:** | Context | Recommended Angle | |---|---| | Recent funding round | Congrats + growth/scale challenge framing, budget window open | | New executive hire | Fresh start, new vendor evaluation cycle, open to change | | Recent product launch | Integration opportunity, complementary capability | | Cost pressure signal | ROI-first framing, fast time-to-value, efficiency story | | Competitor win in their space | Urgency angle, competitive pressure, fear of falling behind | | Award / milestone | Validation: "how do you maintain this standard at scale?" | | Pain signal from job postings | Precision hook: "we saw your [X] postings, we solve that" | **Personalization Layers (stack all four for highest conversion):** 1. **Company layer** — Reference specific company news, milestone, or challenge [EXPLICIT] 2. **Role layer** — Speak to their specific responsibilities and KPIs [INFERRED] 3. **Person layer** — Reference their public content, career moves, or interests [EXPLICIT] 4. **Timing layer** — Connect to a trigger event (funding, launch, Q1 planning cycle) [INFERRED] **Channel Sequence Recommendation:** 1. LinkedIn connection request (no note, or short note referencing shared context) 2. LinkedIn message after connection (1 paragraph, specific hook, one clear ask) 3. Email follow-up 5 days later (reference the LinkedIn outreach, add new value element) 4. Warm intro via mutual connection if one exists (highest conversion, always preferred) **Objection Pre-empts:** | Common Objection | Pre-empt in the Message | |---|---| | "We already have a solution" | Acknowledge + differentiate on one specific dimension | | "Not the right time / budget frozen" | Ask for a 15-min discovery, not a commitment | | "Send me more info" | Offer a 1-pager tailored to their use case, not a generic deck | | "We handle this internally" | Validate their approach, offer a benchmark/audit angle | | "We're in evaluation mode already" | Ask to be included, offer a specific POV or framework | **Opening Message Draft Formula:** ``` [Trigger event / personalization hook] + [specific pain we solve] + [1 reference to a peer / analogous company] + [metric or outcome] + [single soft ask] ``` Example: > "Saw your team posted 6 data engineer roles this quarter — teams scaling data infra > at your stage usually hit the orchestration wall 60 days after the dbt rollout. > We helped [Peer Company] cut pipeline failures by 70% in that window. > Worth a 20-minute call to see if the pattern fits?" --- ## S6: HTML Dossier Output Template Generate a branded, self-contained HTML dossier. Load full template from `references/html-template.md`. [EXPLICIT] **Brand tokens**: `--navy: #122562` · `--gold: #FFD700` · `--blue: #137DC5` · dark bg `#0d0d0d` · Poppins headings + Inter body. **Sections in the HTML output**: Header (company name + date + confidential badge) → TL;DR executive brief (3 bullets) → Company DNA (10-field data grid) → Key Contacts (2 contact cards) → Pain Hypotheses (3 items with confidence badges) → Approach Strategy (4-step numbered list) → Footer (JM Labs attribution). **Delivery rules** [EXPLICIT]: - Replace every `[placeholder]` with researched data - Label inferred fields with `[est.]` inline - Do not deliver with unfilled placeholders - Mark `[INFERRED]` on estimated revenue / headcount / tech stack when source is indirect --- ## Trade-off Matrix | Decision | Option A | Option B | Recommendation | |---|---|---|---| | Depth vs. speed | Full 6-section dossier (2-3h) | Quick 2-section brief (30min) | Full for strategic accounts; brief for same-day inbound | | Public vs. inferred data | Only confirmed facts | Include reasoned inferences | Include inferences, tag clearly with [INFERRED] | | HTML vs. markdown output | Branded HTML dossier | Plain markdown summary | HTML for client-facing; markdown for internal notes | | Company vs. person focus | Company-first, then people | Person-first, then company | Company-first unless user specifies a named target | | One contact vs. full map | Single decision-maker | Full stakeholder map | Full map for deals >$50K; single for fast outreach | | Hypothesis depth | 1 primary pain | 3–5 ranked hypotheses | 3–5 always; more hypotheses = better discovery call prep | --- ## Assumptions & Limits - [EXPLICIT] All research is based on publicly available information only. No paid databases, no data brokers, no private data inference beyond professional context. - [INFERRED] Revenue and headcount figures for private companies are estimates. Every such figure is labeled `[est. — not verified]` in the output. - [EXPLICIT] Email patterns are hypotheses, not confirmed addresses. User must validate before use. - [OPEN] Tech stack inferences from job postings may lag reality by 6–12 months. Directional only. - [EXPLICIT] This skill does not profile private individuals outside a professional or public-figure context. If a request concerns a private person, decline and explain the privacy concern. - [INFERRED] Pain hypotheses are probabilistic. The validation question in each hypothesis is mandatory — do not treat as confirmed until the discovery call verifies it. - [EXPLICIT] Dossier freshness degrades fast — refresh if >30 days old before any meeting. --- ## Edge Cases **EC-1: Company with no digital footprint** Very small or recently founded company has minimal LinkedIn presence, no press, minimal web. Action: Build a lightweight dossier from whatever is available. Flag the confidence score as low. Use industry + company size + founding year to construct pain hypotheses from first principles (e.g., "at 20 employees in SaaS, the most common pain at 18 months is [X]"). **EC-2: Executive with a common name / ambiguous identity** "John Smith at Acme" — multiple LinkedIn matches, ambiguous results. Action: Ask user for LinkedIn URL or full title before proceeding. Do not produce a dossier on the wrong person. Verify company + current title before writing any Contact Card. **EC-3: Company in a language other than English** Target is a Latin American or European company with minimal English-language presence. Action: Research in the native language. Produce the dossier in the language the user requests. Use `[data-l="es"]` / `[data-l="en"]` HTML markup throughout the output. **EC-4: User wants a dossier on a competitor (not a prospect)** Intent is competitive intelligence, not a sales dossier. Action: Redirect to the competitive-intelligence skill. The dossier framework is optimized for outreach and entry strategy, not competitive positioning — wrong tool for this use case. **EC-5: Publicly traded company with overwhelming data volume** 10-K, earnings calls, IR page, analyst coverage — information overload. Action: Prioritize recency (last 12 months) and strategic relevance (what affects this deal). Summarize, do not dump. The dossier is a decision tool, not an archive. Cite specific sections for the user to read if they want more depth. --- ## Good vs. Bad Example **BAD — Generic, no action value:** > "Acme Corp is a technology company with around 500 employees. They seem to be growing > and might be interested in our services. Their CEO is John Doe." **GOOD — Specific, hypothesis-driven, actionable:** > "Acme Corp (Series B, $42M raised Jan 2024 [EXPLICIT], 340 → 480 employees in 12 months > [INFERRED via LinkedIn]) is scaling engineering faster than their data infrastructure > can support. Evidence: 6 open data engineer roles in Q1 [EXPLICIT], all requiring dbt + > Snowflake, none requiring Airflow — the orchestration gap [INFERRED]. Decision-maker: > Maria Gómez, VP Engineering (joined 8 months ago from Stripe [EXPLICIT], previously built > a data platform team from scratch [EXPLICIT via LinkedIn bio]). Hook: 'We saw your data > engineer postings — teams at your stage typically hit the orchestration wall 60 days after > the dbt rollout. We solved this for [Peer Company] and cut pipeline failures by 70%. > Worth a 20-min call?' [INFERRED pain + EXPLICIT peer reference]." The good version has: named signals with evidence tags, a specific gap, a named contact with career context, a peer reference, a metric, and a specific one-ask close. Every element traces to a real signal source. --- ## Validation Gate Before delivering any dossier output, verify all of the following: - [ ] Company name correctly spelled and verified — not a homonym or alias - [ ] All revenue / headcount figures labeled `[est.]` if not from a public source - [ ] At least 2 specific signals support the primary pain hypothesis - [ ] Every named contact drawn from public professional sources only - [ ] Email patterns labeled as `[INFERRED — not verified]`, not as confirmed addresses - [ ] Tech stack claims trace to at least one concrete signal (job posting, website, press) - [ ] Approach strategy includes a specific personalization hook, not a generic opener - [ ] HTML output uses JM Labs brand tokens (navy `#122562`, gold `#FFD700`, blue `#137DC5`) - [ ] Bilingual `[data-l]` attributes present on all user-facing strings in HTML output - [ ] All `[placeholder]` tokens in HTML have been replaced with real researched data --- ## Reference Files | File | Purpose | |---|---| | `knowledge/body-of-knowledge.md` | Sales intelligence and OSINT methodology | | `knowledge/knowledge-graph.md` | Entity relationships for prospect research | | `evals/` | Scored dossier examples for quality calibration | --- ## Related Skills - `market-intelligence` — Broader market/sector context before the dossier - `b2b-outreach` — Convert dossier findings into outreach sequences - `client-prospecting` — Identify who to build dossiers for - `competitive-intelligence` — Competitor profiles (different framework from prospect dossiers) - `executive-pitch` — Use dossier findings to prepare a pitch narrative - `stakeholder-mapping` — Internal org mapping when target is an existing client