Available Context & Tools

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

Deal Slippage Diagnosis

Goal

Diagnose at-risk deals across the pipeline, identify the specific slippage signals for each, determine root causes, and generate a prioritized rescue plan with concrete actions. This is a pipeline-level diagnostic -- it looks across ALL deals to find the ones that need attention most, ranks them by severity and save-ability, and produces a focused action plan that a rep or manager can execute immediately.

Why Pipeline Slippage Diagnosis Matters

Pipeline slippage is the #1 destroyer of sales forecasts and quota attainment. The data is stark:

57% of forecasted deals slip to the next quarter (Clari, 2023 Revenue Operations Report). The average B2B pipeline is 57% fiction -- deals that will not close when the rep says they will.
Only 28% of reps hit quota. The primary reason is not effort or skill -- it is misallocation. Reps spend time on dead deals while winnable deals slip unnoticed (CSO Insights, World-Class Sales Practices).
Early detection changes everything. Deals that receive intervention within 7 days of showing risk signals have a 3.2x higher recovery rate than deals that go 30+ days before someone notices (InsightSquared pipeline analytics).
The average rep loses 8 hours per week on deals that will never close. That is 400+ hours per year -- enough to work 40 additional winnable deals (Salesforce State of Sales, 2023).
Forecast accuracy improves by 42% when slippage signals are systematically monitored and flagged (Gartner, Sales Analytics Study).

The goal is not to be negative about the pipeline -- it is to be honest. Honest pipelines close more deals because rep time is allocated to deals that can actually be won.

Required Capabilities

CRM: To fetch pipeline deals, deal details, activities, health signals, and contacts
Tasks: To create rescue task previews

Inputs

deal_id (optional): Specific deal to diagnose. If omitted, scan the full pipeline for at-risk deals.
deal_context (optional): Additional context such as health scores or known risk factors.
limit (optional, default 10): Maximum number of at-risk deals to analyze.

Data Gathering (via execute_action)

Full Pipeline Scan (when deal_id is not provided)

Fetch at-risk deals: execute_action("get_pipeline_deals", { filter: "at_risk", include_health: true, limit: 10 })
For each deal, fetch details: execute_action("get_deal", { id, include_health: true })
Query health scores: Fetch from deal_health_scores table for overall_health_score, risk_factors, risk_level, days_in_current_stage, sentiment_trend, meeting_count_last_30_days, days_since_last_activity
Query relationship health: Fetch from relationship_health_scores for primary contacts -- check is_ghost_risk, ghost_probability_percent, days_since_last_contact, response_rate_percent
Fetch recent activities per deal: execute_action("get_deal_activities", { deal_id, limit: 10 })
Fetch overdue tasks: execute_action("list_tasks", { status: "overdue" })
Fetch pipeline summary: execute_action("get_pipeline_summary", {}) -- for context on overall pipeline health

Health data priority: Use deal_health_scores.risk_factors as a primary signal source. This field contains pre-computed risk signals like stage_stall, no_activity, sentiment_decline, no_meetings. Cross-reference these with the signal taxonomy to avoid duplicate detection logic.

Single Deal Diagnosis (when deal_id is provided)

Fetch the deal: execute_action("get_deal", { id: deal_id, include_health: true })
Query health score: Fetch from deal_health_scores table -- this provides pre-computed risk factors, component scores, and trend data
Query relationship health: Fetch from relationship_health_scores for all contacts on the deal -- prioritize by ghost_probability_percent descending
Fetch activities: execute_action("get_deal_activities", { deal_id, limit: 30 })
Fetch tasks: execute_action("list_tasks", { deal_id })
Fetch contacts on deal: from deal record -- check engagement levels

Health score as diagnostic shortcut: The deal_health_scores.risk_factors array and sentiment_trend field provide pre-analyzed signals. Use these to confirm manual signal detection and prioritize diagnosis. For example, if risk_factors includes sentiment_decline and sentiment_trend: 'declining', you have high-confidence signal detection without recalculating from raw meeting data.

If data calls fail or return empty, note the gap and work with available data. Missing data is itself a risk signal.

Slippage Signal Taxonomy

Read references/slippage-signals.md for the deep dive on all 15+ signals with CRM detection queries, historical accuracy rates, false positive indicators, and signal interaction effects. See references/diagnostic-frameworks.md for the 5 Whys methodology, deal review meeting agendas, and manager escalation templates.

These are the 15 most predictive slippage signals, ranked by severity. When analyzing a deal, check for each signal and note which ones are present. The combination of signals determines the overall risk severity.

Critical Severity (deal likely to slip or die without immediate intervention)

Signal 1: No activity in 14+ days

What: No emails, calls, meetings, or notes for 14+ calendar days
Why critical: Gong research shows deals that go dark for 14+ days close at less than 8%. Activity is the oxygen of a deal.
Detection: Compare current date to last activity date on the deal
Common cause: Champion went dark, internal priority shift, competitor won

Signal 2: Close date pushed 2+ times

What: The expected close date has been moved back at least twice
Why critical: Each push reduces close probability by 15-20%. Two pushes means the original forecast was wrong AND the first correction was also wrong. The pattern is predictive of further slippage.
Detection: Check deal history for close date changes. If no history is available, compare current close date to original/created date.
Common cause: No compelling event, overconfident initial forecast, buyer process is longer than expected

Signal 3: Close date is in the past

What: The expected close date has already passed and the deal is still open
Why critical: This is the most obvious slippage signal and the most common. It indicates either the deal was lost and not updated, or the close date was fictional from the start.
Detection: Compare close date to today's date
Common cause: Rep forgot to update CRM, buyer delayed without communication, deal is actually dead

Signal 4: Deal value decreased by 20%+

What: The deal amount was reduced significantly
Why critical: Value reduction signals scope retreat. The buyer is either de-scoping to fit a tighter budget, or the seller is discounting to stay competitive. Either way, the deal dynamics have shifted unfavorably.
Detection: Compare current deal value to original/maximum value
Common cause: Budget constraints, competitive pressure, reduced scope

High Severity (deal will likely slip without action this week)

Signal 5: Single-threaded engagement

What: Only one contact from the buyer's side is engaged
Why critical: Single-threaded deals close at 5.4% vs. 17% for multi-threaded (Gong). If that one contact goes on vacation, changes roles, gets overruled, or deprioritizes -- the deal dies instantly.
Detection: Count unique contacts with activity in the last 30 days
Common cause: Rep did not ask for introductions, champion is gatekeeping, deal is too early-stage for multi-threading

Signal 6: No executive/economic buyer engagement

What: No meeting, email, or call involving the person who controls the budget
Why critical: Deals without economic buyer engagement in the first 60% of the sales cycle close at 11% vs. 34% when the buyer is engaged early (MEDDIC benchmark data).
Detection: Check contact roles on the deal. If no one has a title suggesting budget authority (VP+, Director, Owner, C-level), flag it.
Common cause: Champion is shielding the process, rep is afraid to ask for access, economic buyer does not know the initiative exists

Signal 7: Stage duration exceeds 2x average

What: The deal has been in its current stage for more than twice the average duration for that stage
Why critical: Extended stage duration is the most reliable leading indicator of eventual loss. It means something is blocking progress and it has not been identified or addressed.
Detection: Compare days in current stage to the average stage duration (if available from pipeline data) or use these defaults: Discovery 14 days, Evaluation 21 days, Proposal 14 days, Negotiation 10 days, Closing 7 days
Common cause: Hidden objection, missing stakeholder, technical blocker, buyer indecision

Signal 8: Overdue tasks on the deal

What: Tasks associated with the deal are past their due date
Why critical: Overdue tasks mean commitments were made but not kept -- either by the rep or the buyer. This erodes trust and slows momentum.
Detection: Check task list for tasks with due_date < today and status != complete
Common cause: Rep overcommitted, buyer did not follow through on agreed actions, tasks were not properly tracked

Medium Severity (deal needs attention within 2 weeks)

Signal 9: Activity frequency declining

What: The rate of activity (meetings, emails, calls) is decreasing compared to the previous period
Why medium: Declining activity is a leading indicator of future silence. It is not yet critical, but the trend is moving in the wrong direction.
Detection: Compare activity count in the last 14 days to the 14 days before that
Common cause: Buyer losing interest, competing priorities, evaluation fatigue

Signal 10: No next meeting scheduled

What: There is no upcoming meeting or call on the calendar for this deal
Why medium: A deal without a next meeting is a deal without momentum. The next interaction is uncertain, which means the buyer is not committed to continuing the process.
Detection: Check for future calendar events or tasks with type "meeting" or "call"
Common cause: Rep forgot to book the next meeting at the end of the last one, buyer avoided committing to a next step

Signal 11: Health score below 50

What: The deal's computed health score (from deal_health_scores.overall_health_score) is below the 50% threshold
Why medium: Health scores aggregate multiple signals. Below 50 means multiple factors are combining to create risk.
Detection: Query deal_health_scores table for overall_health_score < 50
Common cause: Multiple contributing factors -- the health score is a composite signal from stage velocity, sentiment, engagement, and activity scores
Cross-reference: Also check deal_health_scores.risk_factors array and risk_level field for specific diagnosis

Signal 12: No mutual action plan or defined next steps

What: There is no documented MAP, and the deal has no structured plan for progressing to close
Why medium: Deals without a MAP close 49% slower and at a 18-22% lower rate. Without structure, both sides drift.
Detection: Check for MAP-related tasks or notes. If none, flag this signal.
Common cause: Rep did not create a MAP, buyer did not agree to structured process

Low Severity (monitor and address proactively)

Signal 13: Buyer company shows organizational change

What: News of layoffs, M&A, leadership change, or restructuring at the buyer's company
Why low (but important): Organizational change does not always kill deals, but it reshuffles priorities and decision-making. It is a risk to monitor, not necessarily to panic about.
Detection: Check deal notes or known intel about the buyer's company
Common cause: External factors unrelated to your deal

Signal 14: Competitor mentioned in recent interactions

What: The buyer has referenced evaluating or considering an alternative solution
Why low (but important): Knowing about a competitor is actually better than not knowing. It gives you a chance to differentiate. The risk is only high if you cannot differentiate.
Detection: Check activity notes and meeting transcripts for competitor mentions
Common cause: Buyer is running a standard evaluation process

Signal 15: Contract/legal review taking longer than expected

What: Legal, security, or procurement review has exceeded the expected timeline
Why low (but escalatable): Legal review delays are common and usually resolvable. They become high-severity if they indicate a fundamental disagreement on terms.
Detection: Check for legal/procurement tasks or milestones that are overdue
Common cause: Legal team is backlogged, terms need negotiation, security review raised questions

Root Cause Analysis Methodology

After identifying the signals present on each deal, determine the root cause using this hierarchy:

If the buyer is not responding (Signals 1, 5, 9): Root cause is likely "stale_engagement" or "champion_dark"
If timeline keeps moving (Signals 2, 3, 7): Root cause is likely "no_compelling_event" or "missing_decision_maker"
If value is decreasing (Signal 4): Root cause is likely "budget_uncertainty" or "competitor_risk"
If execution is failing (Signal 8): Root cause is likely "poor_execution" (internal issue, not buyer issue)
If structure is missing (Signals 10, 12): Root cause is likely "no_process" -- the deal lacks a framework for progressing
If external factors are present (Signals 13, 14, 15): Root cause is likely "external_blocker" -- outside the rep's direct control

Root causes to use in the output:

stale_engagement: Buyer interest has cooled, activity has dropped
champion_dark: Primary contact has stopped responding
missing_decision_maker: Economic buyer not engaged, deal cannot progress
budget_uncertainty: Budget not confirmed, reduced, or reallocated
competitor_risk: Alternative solution gaining traction
procurement_blocker: Legal, security, or procurement process stalling the deal
no_compelling_event: No urgency or deadline driving the buyer to act
poor_execution: Internal failure (overdue tasks, missed follow-ups, no MAP)
organizational_change: Buyer's company undergoing restructuring, M&A, or leadership change

Pipeline Risk Scoring Framework

For pipeline-level analysis, score each deal to determine which ones to focus on first. The goal is to answer: "If I only have time for 3 deals this week, which 3 should I focus on?"

Deal Risk Score Calculation

Risk Score = (Signal Severity Sum) x (Deal Value Weight) x (Close Date Proximity)

Signal Severity Sum: Critical = 10 points, High = 5 points, Medium = 2 points, Low = 1 point. Sum all detected signals.

Deal Value Weight: Normalize deal value to a 1-3 multiplier:

Top 25% of pipeline by value: 3x
Middle 50%: 2x
Bottom 25%: 1x

Close Date Proximity: Deals closer to close date are more urgent:

Close date in next 2 weeks: 3x
Close date in next month: 2x
Close date 1+ months out: 1x
Close date in the past: 3x (already slipped)

Rescue Prioritization

After scoring, rank deals by risk score (highest first). Then apply these triage rules:

Risk Score	Severity	Action	Timeline
50+	Critical	Immediate rescue plan, manager notification	Today
25-49	High	Rescue plan, rep executes this week	This week
10-24	Medium	Proactive intervention, schedule actions	Next 2 weeks
1-9	Low	Monitor, no immediate action needed	Ongoing

The "Save or Kill" Decision: For deals with Risk Score 50+, the first question is not "how do we save this?" but "SHOULD we save this?" A high-risk, low-value deal may not be worth the rescue effort. A high-risk, high-value deal with a salvageable root cause deserves immediate attention.

Rescue Action Prioritization

For each at-risk deal, generate rescue actions ranked by this framework:

Action Priority Rules

Respond to buyer silence first. If the buyer is not responding, nothing else matters. Re-engagement is always the #1 priority for stale deals.
Address the root cause, not the symptoms. If the root cause is "missing decision-maker," do not prescribe "send another email to the champion." Prescribe "request an executive alignment meeting."
Prefer low-effort, high-impact actions. A 10-minute phone call to a champion is almost always higher-ROI than a 2-hour proposal revision.
Switch channels for stale deals. If email has failed, prescribe a call or LinkedIn message. If calls have failed, go through a different contact.
One deal, one primary action. Each deal should have one "do this first" action. A list of 10 actions per deal causes paralysis, not progress.

Action Templates by Root Cause

stale_engagement:

Send a value-add email with a relevant insight or case study (not "just checking in")
Call the champion with a specific question that requires a response
Reach out to a secondary contact for a project status update

champion_dark:

Go around: contact a different stakeholder
Go above: executive-to-executive outreach
Send the breakup email after 72 hours of multi-channel silence

missing_decision_maker:

Ask champion directly: "Who else needs to be involved to move this forward?"
Offer an executive briefing for the economic buyer
Create and share a business case document the champion can use internally

budget_uncertainty:

Build/share the ROI model quantifying cost of inaction
Propose a phased approach or smaller initial scope
Identify alternative budget holders or fiscal year timing

competitor_risk:

Ask directly about the competitive landscape
Share a reference from a customer who evaluated the same competitor
Differentiate on the buyer's specific evaluation criteria (not generic features)

procurement_blocker:

Ask for a specific list of procurement requirements
Offer to complete security questionnaires, vendor registration proactively
Introduce your legal/security team directly to their procurement team

no_compelling_event:

Quantify the cost of delay (monthly cost of the problem)
Introduce a real constraint (capacity, pricing, or market timing)
Find or create an internal deadline (budget cycle, board meeting, audit)

poor_execution:

Clear the overdue task backlog (complete or close each one)
Create a MAP with milestones and exit criteria
Book the next meeting before doing anything else

organizational_change:

Research the new landscape (who is in charge now?)
Reach out to the new decision-maker with a fresh executive summary
Set a reminder to re-engage after the transition stabilizes (30-90 days)

Slack Notification Best Practices

The slack_update_preview should be formatted for quick consumption by a sales manager. Follow these rules:

Format Rules

Lead with the headline: "Pipeline Risk Report: [X] deals need attention"
Use emoji-free, professional formatting (the dashboard does not use emojis)
Bold the deal names and risk levels
Include deal values to help the manager prioritize
Limit to top 3-5 deals (do not overwhelm)
End with a clear ask or recommendation

Content Structure

*Pipeline Risk Report: [X] deals flagged*

*[Deal Name]* ($[Value]) - [Severity]
  Risk: [Primary signal in plain language]
  Action: [Top rescue action in one sentence]

*[Deal Name]* ($[Value]) - [Severity]
  Risk: [Primary signal]
  Action: [Top rescue action]

*Recommended*: [1-2 sentence summary of what the manager should know or do]

When to Include Slack Notification

Always include when 2+ deals are at Critical or High severity
Include when total at-risk pipeline value exceeds 20% of the rep's quota
Include when a single deal above $100K is at Critical severity
Skip for Low severity only situations -- do not cry wolf

Manager Escalation Criteria

Flag for manager attention when any of these conditions are met:

Condition	Escalation Level	Rationale
Deal $200K+ at Critical severity	Immediate	High-value deal at high risk requires executive attention
3+ deals at High severity for same rep	This week	Pattern suggests rep capacity or skill issue
Close date pushed 3+ times	Immediate	Forecast is unreliable, need honest assessment
Single-threaded $100K+ deal	This week	Structural risk that the rep may need help addressing
Competitor winning and deal $150K+	Immediate	May need executive intervention or strategic response
Rep has 50%+ pipeline at risk	This week	Quota is in jeopardy, may need pipeline generation support

Manager escalation should be supportive, not punitive. The Slack message should frame risks as "opportunities to help" not "problems to investigate."

Output Contract

Return a SkillResult with:

data.risk_radar: array of 5-8 at-risk deals (sorted by risk score, highest first)
- deal_id: string
- deal_name: string
- company: string | null
- value: number | null
- close_date: string | null
- days_since_last_activity: number | null
- risk_score: number (calculated from the risk scoring framework)
- risk_signals: string[] (specific signals detected, using taxonomy names)
- root_cause: string (from the root cause categories)
- severity: "critical" | "high" | "medium"
- salvageable: boolean (honest assessment)
- one_line_diagnosis: string (plain-language summary of why this deal is at risk)
data.rescue_actions: array of 5-8 rescue actions (ranked by priority across all deals)
- title: string (specific, includes deal name and action)
- description: string (detailed enough to execute without follow-up questions)
- deal_id: string
- deal_name: string
- priority: "urgent" | "high" | "medium"
- estimated_time: number (minutes)
- roi_rationale: string (why this action matters for THIS deal)
- root_cause_addressed: string (which root cause this action targets)
data.task_previews: array of 3 task previews (the most important rescue actions, ready to create as tasks)
- title: string (specific, includes person/deliverable)
- description: string (include checklist of subtasks)
- due_date: string (ISO date, prefer "today" or "tomorrow" for urgent items)
- priority: "high" | "medium" | "low"
- deal_id: string
data.slack_update_preview: object (for manager notification)
- channel: "slack"
- message: string (Slack-formatted summary -- see Slack Notification Best Practices)
- blocks: optional Slack Block Kit payload (for richer formatting)
- should_send: boolean (based on escalation criteria -- do not send for low-risk-only situations)
- escalation_reason: string | null (why this warrants manager attention)

Quality Checklist

Before returning the diagnosis, verify:

Examples

Good Risk Radar Entry

{
  "deal_id": "deal_abc123",
  "deal_name": "Acme Corp - Platform License",
  "company": "Acme Corp",
  "value": 185000,
  "close_date": "2026-02-28",
  "days_since_last_activity": 19,
  "risk_score": 72,
  "risk_signals": [
    "no_activity_14_days",
    "single_threaded",
    "close_date_pushed_twice",
    "no_economic_buyer_engagement"
  ],
  "root_cause": "missing_decision_maker",
  "severity": "critical",
  "salvageable": true,
  "one_line_diagnosis": "Only contact (Sarah Chen, Director) went dark 19 days ago. VP-level buyer never engaged. Close date pushed twice from Jan 15 to Feb 28. Without executive involvement, this $185K deal will not close."
}

Why this works: Specific signals cited with data. Root cause is precise. One-line diagnosis tells the full story in one sentence with names, dates, and amounts.

Bad Risk Radar Entry

{
  "deal_id": "deal_abc123",
  "deal_name": "Acme Corp",
  "company": "Acme Corp",
  "value": 185000,
  "close_date": "2026-02-28",
  "days_since_last_activity": null,
  "risk_score": null,
  "risk_signals": ["at_risk"],
  "root_cause": "general_risk",
  "severity": "high",
  "salvageable": true,
  "one_line_diagnosis": "This deal is at risk and needs attention."
}

Why this fails: No specific signals. No risk score. Root cause is meaningless. Diagnosis says nothing actionable.

Good Slack Notification

*Pipeline Risk Report: 3 deals flagged this week*

*Acme Corp - Platform License* ($185K) - CRITICAL
  Risk: No activity in 19 days, VP buyer never engaged, close date pushed twice
  Action: Request executive alignment meeting through Sarah Chen's VP of Engineering

*GlobalTech - Analytics Suite* ($92K) - HIGH
  Risk: Single-threaded through a mid-level PM, close date is Feb 21
  Action: Ask PM to introduce you to the Director of Data before the close date

*StartupCo - Starter Plan* ($28K) - MEDIUM
  Risk: Demo completed 3 weeks ago with no follow-up, no next meeting scheduled
  Action: Send ROI model with cost-of-delay analysis, propose decision meeting

*At-risk pipeline total: $305K (32% of Q1 target)*
Recommendation: Acme Corp needs executive-to-executive outreach this week -- it is the highest-value deal at critical risk.

Error Handling

No at-risk deals found

If the pipeline scan returns no at-risk deals, this is GOOD news. Report it clearly: "No deals currently show slippage signals. Pipeline health is strong." Then provide proactive recommendations: "To maintain this, ensure every deal has: a next meeting scheduled, multi-threaded engagement, and a defined MAP."

Pipeline data is incomplete or unavailable

If get_pipeline_deals fails or returns minimal data, fall back to available information. If only a single deal_id was provided, diagnose that deal thoroughly. If no deal data is available at all, explain what data is needed and how to ensure CRM records are complete.

Too many at-risk deals (15+)

If the pipeline has 15+ at-risk deals, do not analyze all of them. Apply the risk scoring framework to rank them, then focus on the top 5-8. Note: "Your pipeline has [X] deals showing risk signals. Analyzing the top [Y] by risk score. Consider reviewing the remaining [Z] deals for potential disqualification -- a smaller, healthier pipeline outperforms a large, leaky one."

All deals are at-risk

If every deal in the pipeline is flagged, the problem is systemic, not deal-specific. Shift the diagnosis from individual deal rescue to pipeline health: "All [X] deals show risk signals. This suggests a systemic issue rather than individual deal problems. Common causes: (1) Over-qualification at the top of funnel (deals entering pipeline too early), (2) Inconsistent follow-up cadence, (3) Lack of multi-threading, (4) Missing MAPs. Recommend a pipeline review meeting with your manager."

Deal has no close date

Assign the signal "missing_close_date" (Medium severity) and estimate a close date based on stage and deal size. Note: "No close date set. Estimated [date] based on stage and deal value. The first rescue action should be confirming a realistic close date with the buyer."

Mixed signals (positive and negative)

When a deal has both positive signals (recent meeting, increasing engagement) and negative signals (close date pushed, missing economic buyer), report both honestly. Do not let positive signals mask structural risks. "This deal has recent momentum (meeting last week, positive feedback) but structural risk remains: no economic buyer engaged and close date has been pushed twice. The momentum is real but fragile."

Health score is unavailable

If the deal does not have a computed health score, calculate an approximate one from available signals: count the number and severity of detected signals and use that as a proxy. Note: "Health score is not available for this deal. Based on detected signals, estimated risk level is [severity]."

Tone and Presentation

Be diagnostic, not alarmist. "This deal has 3 critical risk signals" not "THIS DEAL IS IN TROUBLE."
Use data and specifics. "No activity in 19 days, close date pushed from Jan 15 to Feb 28" not "the deal is stalling."
Be honest about probability. If a deal is likely dead, say so. False optimism wastes the rep's most scarce resource: time.
Frame rescue actions as the next 72 hours, not a multi-week project. Urgency matters.
The Slack notification should be written for a sales manager who has 30 seconds to read it. Lead with the most important information.
When recommending disqualification, frame it positively: "Closing this deal as lost frees up capacity for [X other winnable deals]. The best pipeline management is knowing which deals to remove, not just which ones to add."
Never use "just" in rescue actions. "Just follow up" minimizes the craft of sales. Every action should have a specific rationale.