ceo-orchestration 1.0.0

package/.claude/skills/domains/sales/skills/outbound-strategist/SKILL.md

@@ -0,0 +1,262 @@
+---
+name: outbound-strategist
+description: >
+  Governs signal-based outbound prospecting: ICP definition with falsifiable exclusion
+  criteria, trigger-signal taxonomy ranked by intent strength and decay window, multi-channel
+  sequence architecture matched to buyer persona, and personalization tiering that separates
+  deeply researched effort from broadcast automation. Use when pipeline sourcing requires
+  a new outbound motion, when reply rates fall below signal-based benchmarks, when an SDR
+  team is measured on send volume rather than pipeline quality, when sequence design needs
+  channel-persona alignment, or when ICP definitions are too broad to generate a targeted
+  account list.
+owner: Sasha Marlowe (Outbound Strategist, sales domain)
+tier: domain:sales
+scope_tags:
+  - outbound-prospecting
+  - signal-based
+  - sequence-design
+  - icp-definition
+  - personalization
+  - multi-channel
+inspired_by:
+  - source: msitarzewski/agency-agents/sales/sales-outbound-strategist.md@783f6a72bfd7f3135700ac273c619d92821b419a
+    license: MIT
+    relationship: structural_inspiration
+    authored_by: ceo-orchestration framework
+    authored_at: 2026-05-07
+# --- smart-loading fields (PLAN-083 Wave 0b sub-agent 0.7c) ---
+domain: sales
+priority: 8
+risk_class: low
+stack: []
+context_budget_tokens: 500
+inactive_but_retained: true
+repo_profile_binding:
+  frontend: {active: false, priority: 10}
+  engine: {active: false, priority: 10}
+  fintech: {active: false, priority: 10}
+  trading-readonly: {active: false, priority: 10}
+  generic: {active: false, priority: 10}
+activation_triggers: []
+# --- K1 paths: native file-touch activation (PLAN-135 W3 unit k1a) ---
+paths:
+  - "**/outbound/**"
+  - "**/sequences/**"
+  - "**/icp/**"
+  - "**/prospecting/**"
+---
+
+# Outbound Strategist
+
+## Cardinal Rule
+
+An outbound message that does not cite a specific signal observed at THIS prospect THIS
+quarter is spam wearing a name. The obligation before any sequence enrollment is to identify
+one observable event at that account — a leadership change, a funding close, a hiring spike,
+a technology shift, a competitor contract expiry — that creates a plausible reason the buyer
+should care about this message now, not next quarter.
+
+## Fail-Fast Rule
+
+Stop and rebuild the motion when three conditions simultaneously hold: (1) reply rate
+sustained below 5% across two full sequence cohorts on signal-enrolled accounts, (2) ICP
+definition cannot exclude at least 40% of accounts in the target list on firmographic or
+behavioral grounds, and (3) no documented signal source is feeding enrollment. Volume is
+not a substitute for relevance. A broken motion run louder produces unsubscribe rates and
+domain reputation damage, not pipeline.
+
+## When to Apply
+
+- Designing or auditing an outbound motion from first principles.
+- Sequence reply rate drops below 12% on signal-enrolled accounts — diagnostic required.
+- SDR team is reporting activity metrics (emails sent, dials made) rather than pipeline metrics.
+- ICP definition review: if the account list exceeds a manageable research threshold without
+  tier segmentation, the ICP is too broad.
+- Channel-persona mismatch: a single channel used uniformly regardless of buyer seniority.
+- Post-funding or post-launch: new signal sources and new buying personas require sequence reset.
+
+## ICP Definition Discipline
+
+A useful ICP is falsifiable. An ICP that does not exclude companies is a TAM slide. Required
+dimensions for a working ICP definition:
+
+**Firmographic**
+- Industry verticals: 2-4 specific named categories, not broad descriptors.
+- Revenue band or employee count range with explicit upper and lower bounds.
+- Geography scope, if go-to-market is regionally constrained.
+- Funding stage or ownership type, if relevant to budget authority or procurement process.
+
+**Technographic**
+- Prerequisite technologies that must already be present for the product to deliver value.
+- Technologies that signal readiness for an adjacent solution or displacement of an incumbent.
+- Stack indicators visible via job postings, vendor documentation, or public data sources.
+
+**Behavioral**
+- The business event that creates urgency in this quarter, not in general.
+- The pain the product resolves that cannot be deferred without measurable cost.
+- The internal stakeholder who owns the pain acutely enough to champion a purchase.
+- The current workaround and its documented cost or failure rate.
+
+**Signal-Density Threshold**
+- Tier 1 sub-segment: accounts with two or more independent signals active simultaneously.
+- Tier 2 sub-segment: accounts with one strong signal or two weak signals.
+- Tier 3: ICP-fit accounts with no current signal — automation only, no manual research spend.
+- Explicit disqualifiers: industries where win rate is documented below 15%; company stages
+  where the product is architecturally premature or commercially overkill.
+
+## Trigger Signals Taxonomy
+
+Signal classification by intent strength, source, decay window, and actionability.
+
+| Signal Class | Representative Sources | Decay Window | Actionability |
+|---|---|---|---|
+| Direct buyer intent | G2 / review site visits, pricing page views, competitor comparison search, RFP announcement | < 24 hours | Route to rep immediately; highest-priority sequence enrollment |
+| Leadership change | LinkedIn, press releases, 10-K, board announcements | 30 days | New executive = new priorities; outreach before incumbent vendor relationship resets |
+| Funding event | Crunchbase, PitchBook, press release | 14 days | Series B+ with stated growth goals signals budget and urgency; tie message to stated use of funds |
+| Hiring spike | LinkedIn, Greenhouse, Lever, job aggregators | 21 days | Department headcount growth signals scaling pain; message to the team owner, not HR |
+| Technology stack shift | BuiltWith, Wappalyzer, job posting language, vendor press release | 45 days | Stack change creates integration requirement or incumbent displacement window |
+| M&A or restructure | SEC filing, press, industry news | 14 days | Integration pressure and tool consolidation create active evaluation cycles |
+| Conference or content signal | Event speaker lists, webinar registrations, whitepaper downloads | 7 days | Recency matters; message must reference the specific content or event, not the category |
+| Competitor contract renewal | Renewal date databases, G2 reviews, champion intelligence | 60 days | Longest decay window; requires champion-sourced timing data to act with precision |
+
+Signal routing rule: signals must reach the assigned rep within 30 minutes of capture.
+After 24 hours, the conversion rate on the signal drops significantly. After 72 hours,
+a competitor has in most documented cases already initiated contact.
+
+## Sequence Architecture
+
+Sequence design is determined by account tier and buyer persona before any other variable.
+Channel mix, cadence rhythm, and touch count derive from those inputs, not from platform
+defaults.
+
+**Channel-Persona Alignment**
+
+| Persona Level | Primary Channel | Secondary | Tertiary |
+|---|---|---|---|
+| C-Suite | LinkedIn (InMail or connection with note) | Warm introduction / referral | Short direct email |
+| VP-level | Email | LinkedIn | Phone with voicemail |
+| Director | Email | Phone with voicemail | LinkedIn |
+| Manager / Individual Contributor | Email | Short-form video (Loom) | LinkedIn |
+| Technical buyer | Email (technical framing) | Community or Slack | LinkedIn |
+
+**Cadence Rhythm (Standard 10-Touch, 28-Day Frame)**
+
+```
+Touch 1  — Day 1,  Email:      Signal-based opening + single value claim + low-friction CTA
+Touch 2  — Day 3,  LinkedIn:   Connection request with personalized note; no pitch
+Touch 3  — Day 5,  Email:      Relevant data point or industry insight tied to their situation
+Touch 4  — Day 8,  Phone:      Call with voicemail referencing previous email by subject line
+Touch 5  — Day 10, LinkedIn:   Engage with their published content or share relevant material
+Touch 6  — Day 14, Email:      Peer case study from a comparable account and situation
+Touch 7  — Day 17, Video:      60-second Loom referencing something specific to their account
+Touch 8  — Day 21, Email:      Different angle — second pain point or adjacent stakeholder view
+Touch 9  — Day 24, Phone:      Final call attempt
+Touch 10 — Day 28, Email:      Breakup email — honest, brief, door left open without pressure
+```
+
+Each touch adds a new value angle. A repeated ask with different wording is not a new touch;
+it is a noise event. Reply-then-pivot rule: any positive reply exits the automated sequence
+immediately and transfers to human-directed follow-up within one business hour.
+
+**Cold Email Anatomy**
+
+```
+SUBJECT LINE
+- 3-5 words, sentence case, reads like an internal thread
+- References signal or account-specific detail, not category or solution name
+- Never all-caps, never punctuation-spam, never vague urgency
+
+OPENING LINE
+- Signal-specific, not a pleasantry or a self-introduction
+- Correct: "Saw the Series B close last week — growth-stage
+  scaling usually surfaces the tooling gaps the previous
+  architecture was not designed for."
+- Incorrect: "I hope this finds you well."
+- Incorrect: "I'm reaching out because we help companies like yours..."
+
+VALUE CLAIM
+- One sentence connecting their current observable situation to a concrete outcome
+- Buyer's vocabulary, not marketing copy
+- Specificity over cleverness: numbers, timeframes, documented outcomes
+
+SOCIAL PROOF (conditional)
+- One line only; include if directly analogous to buyer's situation and verifiable
+- Omit if not genuinely parallel — unearned social proof erodes credibility faster
+  than no social proof
+
+CTA
+- Single, unambiguous, low friction
+- Correct: "Worth 15 minutes to see if this applies to your team?"
+- Incorrect: "I'd love to schedule a 30-minute demo to walk you through..."
+```
+
+## Personalization Tier Matrix
+
+Personalization investment must match account tier and signal strength. Claiming Tier 1
+effort on a Tier 3 message is the defining characteristic of fake personalization — buyers
+recognize it immediately and it is worse than no personalization.
+
+| Tier | Account Definition | Personalization Depth | Research Investment |
+|---|---|---|---|
+| Tier 1 — Deep | Top 50-100 accounts; two or more active signals | Account-specific: annual report language, earnings call themes, stated strategic initiatives, named stakeholder references | 45-90 minutes per account; dedicated rep ownership; multi-thread across 3-5 contacts |
+| Tier 2 — Templated-with-variable | Next 200-500 accounts; one strong signal or ICP-fit with two weak signals | Industry-specific messaging + account-level opening line variable; persona-matched value claim | 10-15 minutes per account; signal-triggered enrollment; 2-3 contacts per account |
+| Tier 3 — Broadcast | Remaining ICP-fit accounts; no current signal | Role and industry tokens only; no account-level research | Automated enrollment; no manual research spend; single primary contact; automated scoring surfaces for promotion |
+
+Tier promotion criteria: a Tier 3 account that generates positive engagement (reply, content
+engagement, website revisit) or acquires a new Tier 1 signal upgrades to Tier 2 automatically
+within the next enrollment cycle. Quarterly review demotes accounts with no signal activity.
+
+## Performance Diagnostics
+
+Leading and lagging metrics are not interchangeable. Leading metrics diagnose the motion
+in flight; lagging metrics confirm outcomes already in the system. Corrective action
+requires leading-metric visibility.
+
+| Metric | Type | What It Diagnoses | Reference Range |
+|---|---|---|---|
+| Signal-to-contact latency | Leading | Routing speed from signal capture to first touch | < 30 minutes |
+| Reply rate (signal-enrolled) | Leading | Message relevance and opening-line quality | 12-25% |
+| Positive reply rate | Leading | Actual interest generated; ICP fit quality | 5-10% |
+| Meeting set rate (reply-to-meeting) | Leading | CTA clarity and objection handling in follow-up | 40-60% of positive replies |
+| Meeting held rate | Lagging | Prospect commitment quality; confirmation and prep quality | 80%+ of meetings set |
+| Stage 1 to Stage 2 conversion | Lagging | Discovery qualification depth; meeting quality | 50%+ |
+| SQL conversion | Lagging | Full-funnel ICP accuracy and sequence targeting | Varies by ACV and market |
+| Sequence completion rate | Process | Rep execution discipline | 80%+ sequences run to completion |
+| Channel mix effectiveness | Process | Which channel produces reply per persona | Review monthly per persona |
+
+Corrective action thresholds: reply rate below 5% on signal-enrolled cohorts triggers
+sequence audit. Stage 1 to Stage 2 below 35% triggers ICP review or discovery coaching
+escalation, not sequence volume increase.
+
+## Anti-patterns
+
+| Anti-pattern | Failure Mode |
+|---|---|
+| Spray-and-pray | Treats outreach as a numbers game; destroys domain reputation, triggers inbox filters, and produces pipeline that does not convert because ICP fit was never enforced |
+| Fake personalization | Opening line inserts a company name or role title without account-specific observation; buyers recognize it; trust cost exceeds any engagement gain |
+| Stale signal enrollment | Acting on a signal more than the published decay window after capture; competitor has already had the conversation; message has no urgency basis |
+| Channel-persona mismatch | Running phone-primary sequences on C-Suite contacts or email-only on Director-and-below; channel preference is a persona property, not a platform default |
+| Volume escalation as fix | Increasing send volume in response to low reply rate; amplifies the broken motion, accelerates domain damage, and produces false-positive meeting data |
+| Automating Tier 1 outreach | Tier 1 accounts have sufficient deal value to justify manual research; automating them signals to the buyer that their account was not worth research investment |
+| Single-variable optimization | Changing subject line, opening line, and CTA simultaneously in a test cohort; three variables changed produce no learnable signal; test one variable per cohort |
+| Sequence abandonment before touch 6 | Most replies come after touch 4; reps who abandon sequences after 2-3 touches report artificially low reply rates and miss late-breaking signals |
+| Vanity metric reporting | Reporting emails sent, dials made, or sequences enrolled without reply rate, positive reply rate, and meeting held rate; activity metrics do not predict pipeline |
+| Ignoring opt-outs | Non-compliance with opt-out requests is a legal and reputational liability; unsubscribe handling must be immediate and complete with no re-enrollment path |
+
+## Cross-References
+
+- `domains/sales/skills/discovery-coach` — call structure and question sequencing for
+  converting outbound-sourced meetings into qualified opportunities.
+- `domains/sales/skills/account-strategist` — multi-stakeholder mapping and account planning
+  for Tier 1 accounts requiring coordinated outbound across multiple contacts.
+- `domains/sales/skills/pipeline-analyst` — pipeline quality analysis and stage-conversion
+  diagnostics when outbound-sourced pipeline underperforms relative to inbound.
+
+## ADR Anchors
+
+- **ADR-058** (Brainstorm Gate and Two-Pass Review): outbound sequence design exemplifies
+  the two-pass discipline — pass one is ICP definition and signal research without drafting any
+  copy; pass two is sequence construction using only what the research surfaced. Collapsing
+  both passes into simultaneous research-and-write produces generic copy that does not reflect
+  specific findings, the same failure mode as reviewers who generate and evaluate in a single
+  motion.

package/.claude/skills/domains/sales/skills/pipeline-analyst/SKILL.md

@@ -0,0 +1,317 @@
+---
+name: pipeline-analyst
+description: >
+  Revenue operations pipeline analysis discipline covering health diagnostics,
+  deal velocity mathematics, forecast accuracy methodology, and data-driven
+  sales coaching from CRM data. Applies structured qualification depth
+  (MEDDPICC), stage-level conversion benchmarks, and engagement-signal
+  adjustment to produce probability-weighted forecasts with explicit confidence
+  intervals — not single-number guesses. Use when: reviewing pipeline coverage
+  against quota, diagnosing stalled deals, constructing a Commit / Best Case /
+  Upside forecast, identifying which stage is the primary conversion bottleneck,
+  evaluating CRM data hygiene before a forecast call, or coaching a rep on
+  deal-level qualification gaps.
+owner: Marcus Oliveira (Pipeline Analyst, domain persona)
+tier: domain:sales
+scope_tags: [revenue-operations, pipeline-health, deal-velocity, forecast-accuracy, sales-analytics, crm-data]
+inspired_by:
+  - source: msitarzewski/agency-agents/sales/sales-pipeline-analyst.md@783f6a72bfd7f3135700ac273c619d92821b419a
+    license: MIT
+    relationship: structural_inspiration
+    authored_by: ceo-orchestration framework
+    authored_at: 2026-05-07
+# --- smart-loading fields (PLAN-083 Wave 0b sub-agent 0.7c) ---
+domain: sales
+priority: 8
+risk_class: low
+stack: []
+context_budget_tokens: 500
+inactive_but_retained: true
+repo_profile_binding:
+  frontend: {active: false, priority: 10}
+  engine: {active: false, priority: 10}
+  fintech: {active: false, priority: 10}
+  trading-readonly: {active: false, priority: 10}
+  generic: {active: false, priority: 10}
+activation_triggers: []
+# --- K1 paths: native file-touch activation (PLAN-135 W3 unit k1a) ---
+paths:
+  - "**/pipeline/**"
+  - "**/forecasts/**"
+  - "**/crm/**"
+---
+
+# Pipeline Analyst
+
+## Cardinal Rule
+
+A forecast not falsifiable against last-quarter prediction error is opinion;
+opinion does not earn coverage targets. Every forecast output MUST carry an
+explicit confidence band (Commit / Best Case / Upside) AND a stated accuracy
+delta versus the prior period — "Q2 Commit was $4.1M, actual was $3.8M,
+error −7.3%" — before any new number is presented for planning. A single
+point estimate without a confidence interval and a retrospective error
+anchors every downstream plan to false precision and trains the organization
+to ignore the analyst.
+
+---
+
+## Fail-Fast Rule
+
+Pipeline review sessions MUST NOT begin with stage-weighted CRM totals as
+the headline number. Stage probability weights are assigned at deal creation,
+not updated on deal behavior; a deal sitting at Stage 3 for 60 days carries
+the same CRM weight as one that entered Stage 3 yesterday. Before any coverage
+ratio is reported, the following gates MUST pass: (1) all deals with zero
+activity in the past 30 days are flagged and excluded from the quality-adjusted
+count; (2) coverage is computed against remaining quota — not full-year quota —
+for the current period; (3) the Commit category contains only deals with a
+defined close path and no open procurement blocker. If any gate fails, the
+headline number is not ready to present.
+
+---
+
+## When to Apply
+
+Activate this skill when the work involves any of the following:
+
+- Running a pipeline review — weekly, monthly, or quarterly — and producing
+  a Commit / Best Case / Upside forecast output.
+- Diagnosing why a quarter is tracking below or above plan before the
+  quarter closes.
+- Identifying the primary stage-level conversion bottleneck in a funnel.
+- Evaluating whether current pipeline creation rate will sustain the quota
+  target two to three quarters out.
+- Scoring individual deal health using MEDDPICC depth and engagement signals.
+- Auditing CRM data quality before a forecast call or board presentation.
+- Coaching a rep or segment on qualification gaps correlated with deal loss.
+
+Skip when: the engagement is pre-pipeline lead generation (use
+discovery-coach instead); the task is a closed-won account expansion plan
+(use account-strategist instead); or the request is a deal-specific
+negotiation strategy (use deal-strategist instead).
+
+---
+
+## Pipeline Health Frame
+
+Pipeline health is four distinct measurements that can diverge. Treat each
+independently — a high-coverage, low-velocity pipeline is a different failure
+mode than a low-coverage, high-velocity one.
+
+### Coverage
+
+The ratio of open weighted pipeline to remaining quota for the current period.
+
+Targets by business context:
+
+| Context | Minimum Coverage Ratio |
+|---|---|
+| Mature, predictable market | 3.0x |
+| Growth-stage or new market | 4.0–5.0x |
+| New rep ramping | 5.0x+ |
+
+Coverage is always computed quality-adjusted: deals with no activity in 30+
+days, missing close dates, or fewer than 5 of 8 MEDDPICC fields populated
+are discounted or excluded from the quality-adjusted figure. Raw stage-weighted
+pipeline is reported separately as a reference, never as the primary number.
+
+### Velocity
+
+How quickly revenue moves through the funnel per unit of time. See
+`## Velocity Mathematics` for the formula and lever-pulling order.
+
+### Quality
+
+MEDDPICC population depth per deal, combined with engagement recency and
+stakeholder breadth. A pipeline of 20 stale, underqualified deals is
+worth less than 8 active, well-qualified opportunities at the same nominal
+dollar value. Quality beats quantity in every coverage calculation.
+
+### Movement
+
+Stage progression rate versus benchmark. Deals that have not advanced in
+more than 1.5× the median stage duration for their segment are classified
+as stalled. Stalled deals MUST be flagged for intervention or removal —
+not carried silently in the forecast.
+
+---
+
+## Coverage Discipline
+
+Coverage analysis follows a fixed sequence; shortcuts produce misleading
+conclusions.
+
+1. Establish remaining quota for the period — not full-year quota.
+2. Pull all open deals with stage, amount, close date, last activity date,
+   and MEDDPICC field count.
+3. Compute raw weighted pipeline (stage probability × amount).
+4. Apply quality adjustment: exclude or discount deals failing the three
+   hygiene gates (no activity 30+ days, missing close date, MEDDPICC < 5/8).
+5. Calculate quality-adjusted coverage ratio = quality-adjusted pipeline /
+   remaining quota.
+6. Segment by deal size (e.g., SMB < $25K, Mid-Market $25K–$150K, Enterprise
+   $150K+). Blended averages hide the signal.
+7. Calculate pipeline creation rate (new qualified pipeline added per week)
+   and project forward to determine whether coverage will be adequate when
+   the period closes.
+
+Gap diagnostics:
+
+- Coverage below 2x with eight or fewer weeks remaining: pipeline creation
+  alone cannot close the gap; deal-level intervention and forecast revision
+  are both required.
+- Coverage appears adequate but quality-adjusted ratio is below 2x: pipeline
+  is volume-inflated; remove or reclassify stale deals before the next call.
+- Coverage healthy but creation rate declining: no current-quarter risk, but
+  the following quarter is at risk; signal must be escalated now.
+
+---
+
+## Velocity Mathematics
+
+Pipeline velocity is the single compound metric that captures pipeline health
+as a rate rather than a snapshot.
+
+```
+Pipeline Velocity = (Qualified Opportunities × Average Deal Size × Win Rate)
+                    / Sales Cycle Days
+```
+
+Each variable is a diagnostic lever with a distinct intervention:
+
+| Lever | Declining signal | Primary intervention |
+|---|---|---|
+| Qualified Opportunities | Top-of-funnel shortfall; shows in revenue 2–3 quarters later | Increase sourcing activity; segment by source |
+| Average Deal Size | Discounting pressure or market shift | Segment; check if specific reps are systematically underpricing |
+| Win Rate | Stage-level decay; check stage-by-stage to isolate where deals die | Process fix (systemic) or coaching (individual) depending on distribution |
+| Sales Cycle Days | Lengthening = competitive pressure, larger buying committees, or qualification gaps | Qualify harder at entry; compress paper process |
+
+Lever-pulling order: address the earliest-stage lever first. Declining
+qualified opportunity volume is a 2–3 quarter leading indicator; declining
+win rate at the Evaluation stage is a current-quarter indicator. Both matter,
+but the order of intervention reflects urgency versus structural impact.
+
+Win rate MUST be segmented before drawing conclusions. An aggregate win rate
+improvement that is driven entirely by closing smaller, lower-competition
+deals is not an improvement signal — it is a mix shift that will suppress
+average deal size in subsequent periods.
+
+---
+
+## Forecast Accuracy Methodology
+
+Forecast categories carry defined confidence thresholds, not subjective labels.
+
+| Category | Confidence Threshold | Inclusion Criteria |
+|---|---|---|
+| Commit | > 90% | Verbal or written agreement; no open procurement blocker; close-date confirmed by buyer |
+| Best Case | 60–90% | Commit + high-velocity qualified deals with active multi-threaded engagement |
+| Upside | < 60% | Best Case + early-stage high-potential deals; explicitly speculative |
+
+Forecast construction steps:
+
+1. Start from historical base rates: what percentage of deals at each stage,
+   in each segment, in the same calendar period one year ago actually closed?
+   These are the base-rate priors. CRM stage probabilities are almost always
+   higher than empirical base rates.
+2. Apply velocity adjustment: deals progressing faster than segment benchmark
+   receive a probability uplift; deals stalled beyond 1.5× median duration
+   receive a discount.
+3. Apply engagement adjustment: multi-threaded deals with buyer-initiated
+   activity in the past 14 days close at materially higher rates than
+   single-threaded, low-activity deals at the same stage.
+4. Apply seasonal adjustment: quarter-end compression, budget cycle timing,
+   and industry-specific patterns are predictable variance — account for them.
+5. Report the delta between this model and the raw stage-weighted CRM total.
+   A large positive delta (model > CRM) suggests deals progressing faster
+   than CRM weights reflect. A large negative delta (model < CRM) is a risk
+   flag requiring deal-by-deal review.
+
+Accuracy tracking is mandatory. Every forecast call records the Commit
+number. After the period closes, the actual versus Commit delta is logged.
+Forecast accuracy cannot improve without this retrospective record.
+
+---
+
+## Stage Definition Discipline
+
+Stage definitions MUST be expressed as buyer-observable entry criteria, not
+as seller activities.
+
+Falsifiable example: "Buyer has completed a live demo and provided written
+evaluation criteria."
+
+Non-falsifiable example: "Demo has been scheduled." — reject.
+
+Criteria checklist for each stage definition:
+
+- Can a third party verify the criterion from CRM data or a meeting record?
+- Does the criterion reflect something the BUYER has done, not the seller?
+- Is there a clear binary: the criterion is met or it is not?
+
+Rotting thresholds: each stage has a maximum age. Deals exceeding 1.5× the
+segment's median stage duration without documented advancement are flagged
+as stalled. Stalled deals are reviewed at the next pipeline call and either
+receive a documented intervention plan or are moved to a later stage in the
+forecast with an appropriate probability discount.
+
+---
+
+## CRM Data Hygiene
+
+Four fields are required for any deal to be included in forecast analysis
+without a documented exception:
+
+| Field | Staleness threshold | Consequence of staleness |
+|---|---|---|
+| Next step with date | 14 days | Deal excluded from Commit; flagged in Best Case |
+| Close date | Must be current period or explicitly deferred | Excluded from period forecast |
+| Amount | Must reflect current scope | Used as-is with a data-quality note |
+| Competitor | Required for late-stage deals | Single-threaded risk flag |
+
+Data hygiene is not an administrative exercise — it is a forecast accuracy
+input. A forecast built on stale CRM data is not a forecast; it is a
+spreadsheet-formatted guess. The analyst's job is to state data quality
+assumptions explicitly before presenting numbers, never to silently
+interpolate missing values.
+
+Field weaponization risk: when CRM fields are used for rep performance
+management rather than pipeline visibility, reps optimise field values for
+manager approval rather than accuracy. Monitor for systematic patterns
+(e.g., every deal shows close date on the last day of the quarter) as a
+signal that fields have been weaponized and can no longer be trusted for
+forecast purposes.
+
+---
+
+## Anti-patterns
+
+| Anti-pattern | Description | Correction |
+|---|---|---|
+| Hopium pipeline | Deals carried in forecast because the rep believes they will close "eventually," not because there is a close plan. | Apply rotting threshold; require documented close path for all Commit-category deals. |
+| Vanity coverage | Reporting raw stage-weighted pipeline as the coverage ratio without quality adjustment. | Always compute quality-adjusted coverage; publish both numbers with explicit delta. |
+| Retrofit forecast | Adjusting the forecast methodology after the quarter closes to explain the miss rather than to improve future accuracy. | Lock forecast methodology at the start of the quarter; post-period review modifies the next-period model only. |
+| Single-stage chokepoint denial | Win rate is declining but the decline is attributed to individual rep performance rather than a systemic process failure at a specific stage. | Segment win-rate decline by stage, rep cohort, and deal size; a systemic drop affecting multiple reps at the same stage is a process failure. |
+| Blended-average blindness | Reporting aggregate metrics across segments, deal sizes, or rep tenure cohorts, hiding the signal in noise. | Always segment before drawing conclusions; blended averages are presented only as a summary after segmented analysis. |
+| Point-estimate anchoring | Presenting a single forecast number without a confidence range, training stakeholders to treat the number as precise. | Enforce Commit / Best Case / Upside at every forecast presentation; single numbers are not accepted outputs. |
+| Last-quarter anchoring | Adjusting this quarter's forecast by incrementing from the prior-quarter actual rather than analyzing from current deal data. | Build forecast from deal-level data up each period; prior-period actual is an accuracy check, not a starting point. |
+
+---
+
+## Cross-References
+
+- `domains/sales/skills/account-strategist` — post-sale expansion pipeline
+  analysis and NRR / GRR tracking.
+- `domains/sales/skills/deal-strategist` — deal-level negotiation and
+  competitive positioning for individual opportunities.
+- `domains/sales/skills/sales-coach` — rep-level performance diagnostics
+  and skill-gap identification correlated with pipeline outcomes.
+
+---
+
+## ADR Anchors
+
+- **ADR-058** — domain skill authoring standards: house-voice constraints,
+  scope_tags schema, and inspired_by attribution requirements that govern
+  this file.