Funnel strategy

Infer and structure the complete sales funnel for a B2B SaaS company based on their GTM motion (PLG, SLG, or hybrid). Produces high-level stage definitions with strategy, leading-indicator inputs, and lagging-indicator outputs for each stage — including post-close stages (retention, expansion, churn prevention). A funnel is not a template you paste; it's an inference from how a company actually acquires, qualifies, and closes customers. This skill reads the company's GTM signals and produces a structured funnel model that other skills consume. Adopt the senior B2B SaaS GTM strategist persona detailed in the premium reference (agent persona section).

When to run

Invoke when user says: "Define the funnel for [company]" · "Sales funnel stages for [company]" · "Pipeline stages for [company]" · "Lead qualification criteria for [company]" · "GTM motion for [company]" · "Funnel strategy for [company]" · "How does [company] acquire customers?" · "What funnel stages should we use?" · "PLG vs SLG for [company]" · "Map the sales process for [company]".

Do NOT invoke when:

• User wants post-signup lifecycle campaigns → /lifecycle-marketing
• User wants content calendar → /content-strategy
• User wants outbound automation workflows → /gtm-engineer
• User wants ICP definition → /icp-behavioural
• User wants competitive analysis → /competitor-research

Auto-suggest after: company-context completes, or user mentions funnel/pipeline/lead-stages/qualification, or user asks about GTM motion / sales process. Always require confirmation before running.

Research substrate: Exa per .claude/rules/exa-protocol.md (auto-loaded). Primary tool web_search_exa for competitor GTM motion + funnel research. Plugin mcp__plugin_exa_exa__web_search_exa preferred; legacy mcp__exa__web_search_exa still mounted. Citations: [VERIFIED: exa_search, {url}, accessed {YYYY-MM-DD}]. Quality gate: ≥3 sources per major claim, ≥50% [VERIFIED], date filter for "recent/latest" claims.

Inputs

Required:

• Company URL — website URL for the target company (user provides).

Recommended (improve quality):

• company-context output — ACV, team size, funding (improves motion detection accuracy)
• icp-behavioural output — persona buying triggers (informs qualification criteria)
• CRM screenshot or pipeline stages — real stages override inferred ones, increases confidence
• ACV or pricing information — directly determines PLG vs SLG likelihood

Validation: Company URL valid + accessible · ACV/team size extracted if company-context available · CRM stages complete if provided.

Steps

Three phases (visual flowchart + per-step narrative in the premium reference).

1. Phase 1.1 — Fetch company website (WebFetch): Homepage primary CTA ("Sign up free" vs "Book a demo"), pricing page (visible plans, self-serve checkout vs "Contact sales"), product pages (signup flow, free trial, freemium), navigation (Docs/API = PLG signal).
2. Phase 1.2 — Run Exa company search (company_research_exa): Company size + stage, industry/category, funding, GTM team composition, market context, competitors.
3. Phase 1.3 — Score signals + classify motion: Apply the signal scoring table from the premium reference. Rules: PLG ≥8 AND SLG ≤3 → PLG · SLG ≥8 AND PLG ≤3 → SLG · Both ≥5 → Hybrid · Else ask user. Cross-reference industry + ACV heuristics.
4. Phase 1.4 — Declare motion with confidence: "Based on [signals], this looks like a [Motion] GTM motion. Key evidence: [2-3 strongest signals]. Confidence: [High/Medium/Low]." If ambiguous, ask user to confirm. Phase 1 checkpoint: motion classified with evidence.
5. Phase 2.1 — Select funnel template from the premium reference: PLG → ONLY PLG stages. SLG → ONLY SLG stages. Hybrid → both tracks with PQL escalation logic. No mixing.
6. Phase 2.2 — Customize each stage: Produce definition (1-2 sentences for THIS company), strategy (specific to product/market/resources), inputs as LEADING indicators with frequency (activities you control: "Publish 3 LinkedIn posts/week"), outputs as LAGGING indicators with target numbers (results you measure: "500 visitors/month"), cadence (weekly/monthly/quarterly), goal logic (work backwards: output target / conversion rate = input volume), owner. Never put metrics in inputs. Tailor to team size, channels, industry norms. Detail in the premium reference.
7. Phase 2.3 — Define qualification criteria: 3-5 signals each for MQL (marketing→sales handoff), SQL (SDR confirms before AE), PQL (product usage signals — PLG/Hybrid only). Use templates from the premium reference, customize.
8. Phase 2.4 — Map Closed Lost re-entry paths: Brief table — Loss Reason / Re-entry Stage / Trigger / Timing.
9. Phase 3 — Post-close mapping (4 stages, same definition+strategy+inputs+outputs format): Onboarding (path to first value — PLG in-app guided vs SLG CSM-driven). Retention (touchpoints, multi-threading, health scoring). Expansion (usage-limit/team-adoption/new-use-case signals → seat upgrade, tier upgrade, add-ons). Churn prevention (declining usage, champion departure, support escalations, renewal-without-engagement, competitive evaluation). Match motion: PLG = product-led; SLG = CSM-driven.
10. Build output document following the canonical scaffold in the premium reference: GTM motion assessment table → pre-close stages → qualification criteria → post-close stages → Closed Lost re-entry → full journey ASCII → FETE mapping (Find/Enrich/Transform/Export → funnel stages → activities) → gaps + recommendations → source appendix.
11. If client has revenue targets — apply backward funnel math from the premium reference (Revenue → Deals needed → Opportunities → SQLs → MQLs → Leads → Traffic). Validate funnel capacity vs growth goals.
12. Run guardrails in the premium reference (8 anti-hallucination rules: never invent stages, don't assume motion, criteria must be specific, mark inferred fields, don't fabricate ACV, post-close matches motion, source everything, don't invent KPI targets).
13. Run quality checklist + 7-test self-eval in the premium reference (CRM-ready, SDR, content strategist, specificity, motion consistency, dashboard, activity tests). If any fails: "Auto-review flagged: [issue]. Want me to fix this or ship as-is?"
14. Present at Review Gate 2 (standard review): GTM motion + full stage map + qualification criteria. Actions: Approve / Adjust / Add stages.
15. Suggest chains post-delivery: gtm-engineer (FETE map) · content-strategy · outreach-emails · sales-enablement (battlecards) · landing-page-copy.

What good looks like

Evaluations

• GTM motion classified with 3+ supporting signals; motion exclusivity respected (no PLG stages in SLG output).
• Every input has a frequency; every output has a target number; every stage has a cadence.
• Goal logic connects input volume to output targets (work-backwards math).
• Inputs are leading indicators (activities); outputs are lagging (metrics) — never confused.
• Qualification criteria are measurable thresholds, not "good fit" platitudes.
• Post-close motion matches pre-close (PLG product-led retention; SLG CSM-driven retention).
• Closed Lost re-entry paths documented; FETE mapping included.
• Source appendix has URL + access date + confidence per signal.
• Stage format consistent across all stages (definition / strategy / inputs / outputs / cadence / goal logic / owner).
• Backward funnel math applied when revenue targets provided.
• All 7 self-eval tests pass: CRM-ready, SDR, content strategist, specificity, motion consistency, dashboard, activity.