make-mp-data 2.1.11 → 3.0.1

package/dungeons/clinch-agi.js

@@ -0,0 +1,632 @@
+
+import dayjs from "dayjs";
+import utc from "dayjs/plugin/utc.js";
+import "dotenv/config";
+import { weighNumRange, pickAWinner, initChance, integer, decimal } from "../lib/utils/utils.js";
+import { uid } from "ak-tools";
+
+const SEED = "clinch-agi-cpo-demo";
+dayjs.extend(utc);
+const chance = initChance(SEED);
+const num_users = 5_000;
+const days = 95;
+
+/** @typedef {import("../types.js").Dungeon} Config */
+
+/**
+ * CLINCH AGI — AI-Native Sales & GTM Workflow Platform
+ *
+ * Models an AI-powered platform where AI agents operate as virtual team members
+ * for sales/GTM teams: joining meetings, summarizing calls, drafting follow-ups,
+ * updating CRM, monitoring pipeline, and coaching reps.
+ *
+ * Built for a CPO interview demo with Uday Chakravarthi (founder of Clinch AGI).
+ * Designed to showcase Mixpanel's analysis features against a realistic B2B SaaS dataset.
+ *
+ * Core product loop:
+ *   Meeting → AI Summary → Follow-up Draft → Send → CRM Auto-Update → Pipeline Insight
+ *
+ * Monetization: Tiered SaaS (starter / growth / enterprise)
+ *
+ * ═══════════════════════════════════════════════════════════════════
+ * HOOKED INSIGHTS — WHAT'S HIDDEN IN THE DATA
+ * ═══════════════════════════════════════════════════════════════════
+ *
+ * 1. THE AI GOT SMARTER — About 60 days ago, meeting summary quality jumped
+ *    dramatically. The old model scored 50-72; the new one hits 75-95+. You
+ *    can see the inflection point clearly in a line chart of quality_score.
+ *
+ * 2. PERSONALIZATION PAYS OFF — Follow-up emails with high personalization
+ *    scores get 3x more replies and clicks than generic ones. The AI's effort
+ *    in crafting personal messages directly drives better sales outcomes.
+ *
+ * 3. AI WORKS WEEKENDS — On Saturdays and Sundays, CRM updates are almost
+ *    entirely autonomous (no human in the loop), and the AI updates more
+ *    fields per record. The agents truly operate as "always-on" team members.
+ *
+ * 4. ENTERPRISE DEALS MOVE FASTER — Enterprise-tier users see deals progress
+ *    through pipeline stages 2x faster than starter-tier users. The platform
+ *    delivers disproportionate value to larger teams.
+ *
+ * 5. INTEGRATIONS PREDICT SUCCESS — Users who connect 5+ integrations are
+ *    tagged as "high" data completeness and overwhelmingly become champions.
+ *    Fewer integrations correlates with at-risk status.
+ *
+ * 6. TIER DRIVES ACTIVATION — Enterprise users convert through every funnel
+ *    at 1.3x the base rate; starter users convert at 0.7x. Onboarding,
+ *    meeting-to-action, pipeline management — all show the same pattern.
+ *
+ * 7. EARLY ADOPTION = RETENTION — Users who deploy 3+ agents and join 5+
+ *    meetings in their first 30 days hit a "power user milestone" and retain
+ *    at dramatically higher rates. Users with zero agent deploys get flagged
+ *    as churn risks — and they do churn.
+ *
+ * 8. BAD FEEDBACK PREDICTS CHURN — Users who give 3+ negative ratings
+ *    (1 or 2 stars) to AI agents show a steep engagement dropoff in the
+ *    second half of their lifecycle. Their event volume falls by ~60%.
+ *
+ * ═══════════════════════════════════════════════════════════════════
+ */
+
+/** @type {Config} */
+const config = {
+	token: "ff821702cde57664bd2b50eb12ead0bf",
+	seed: SEED,
+	numDays: days,
+	numEvents: num_users * 120,
+	numUsers: num_users,
+	hasAnonIds: false,
+	hasSessionIds: true,
+	format: "json",
+	gzip: true,
+	alsoInferFunnels: false,
+	hasLocation: true,
+	hasAndroidDevices: true,
+	hasIOSDevices: true,
+	hasDesktopDevices: true,
+	hasBrowser: false,
+	hasCampaigns: false,
+	isAnonymous: false,
+	hasAdSpend: false,
+	percentUsersBornInDataset: 35,
+	hasAvatar: true,
+	makeChart: false,
+	batchSize: 2_500_000,
+	concurrency: 1,
+	writeToDisk: false,
+	scdProps: {},
+	mirrorProps: {},
+	lookupTables: [],
+
+	// ──────────────────────────────────────────────
+	// EVENTS (18)
+	// ──────────────────────────────────────────────
+	events: [
+		{
+			event: "account created",
+			weight: 0,
+			isFirstEvent: true,
+			properties: {
+				signup_source: ["organic", "referral", "demo_request", "product_hunt", "g2_review"],
+				initial_plan: ["free_trial", "free_trial", "free_trial", "growth", "enterprise"],
+			},
+		},
+		{
+			event: "agent deployed",
+			weight: 3,
+			properties: {
+				agent_type: pickAWinner(["pre_call_researcher", "meeting_assistant", "follow_up_writer", "crm_updater", "pipeline_monitor", "outreach_agent"]),
+				configuration_time_min: weighNumRange(2, 45, 0.4),
+			},
+		},
+		{
+			event: "meeting joined",
+			weight: 5,
+			properties: {
+				meeting_platform: ["zoom", "zoom", "google_meet", "google_meet", "teams", "webex"],
+				meeting_type: ["discovery", "demo", "negotiation", "check_in", "onboarding", "qbr"],
+				participants: weighNumRange(2, 12, 0.5),
+				duration_min: weighNumRange(10, 90, 0.3),
+			},
+		},
+		{
+			event: "meeting summary generated",
+			weight: 5,
+			properties: {
+				summary_length_words: weighNumRange(100, 800, 0.3),
+				action_items_count: weighNumRange(0, 8, 0.5),
+				quality_score: weighNumRange(50, 95, 0.4), // Hook 1 overrides this
+			},
+		},
+		{
+			event: "crm auto-updated",
+			weight: 8,
+			properties: {
+				crm_platform: ["hubspot", "hubspot", "salesforce", "salesforce", "pipedrive"],
+				fields_updated: weighNumRange(1, 6, 0.5),
+				update_type: ["contact_info", "deal_stage", "activity_log", "meeting_notes", "next_steps"],
+			},
+		},
+		{
+			event: "follow-up drafted",
+			weight: 4,
+			properties: {
+				personalization_score: weighNumRange(30, 98, 0.3),
+				template_used: ["post_meeting", "deal_nudge", "re_engagement", "intro", "proposal"],
+				word_count: weighNumRange(50, 400, 0.4),
+			},
+		},
+		{
+			event: "follow-up sent",
+			weight: 3,
+			properties: {
+				channel: ["email", "email", "email", "linkedin", "slack"],
+				personalization_score: weighNumRange(30, 98, 0.3),
+				time_to_send_hr: weighNumRange(0, 48, 0.5),
+				email_outcome: ["pending", "pending", "pending", "opened", "opened", "clicked", "bounced", "replied"], // Hook 2 overrides
+			},
+		},
+		{
+			event: "pipeline alert received",
+			weight: 4,
+			properties: {
+				alert_type: ["deal_stalling", "champion_left", "competitor_mentioned", "no_activity", "budget_risk"],
+				severity: pickAWinner(["low", "medium", "high", "critical"]),
+				deal_value: weighNumRange(5000, 500000, 0.2),
+			},
+		},
+		{
+			event: "deal stage changed",
+			weight: 5,
+			properties: {
+				from_stage: ["lead", "qualified", "proposal", "negotiation", "verbal_commit"],
+				to_stage: ["qualified", "proposal", "negotiation", "verbal_commit", "closed_won", "closed_lost"],
+				days_in_previous_stage: weighNumRange(1, 60, 0.3), // Hook 4 modifies
+			},
+		},
+		{
+			event: "insight generated",
+			weight: 6,
+			properties: {
+				insight_type: pickAWinner(["deal_risk", "buyer_intent", "competitive_intel", "engagement_trend", "revenue_forecast", "coaching_opportunity"]),
+				confidence_score: weighNumRange(55, 98, 0.4),
+				data_sources_used: weighNumRange(1, 5, 0.5),
+			},
+		},
+		{
+			event: "agent feedback",
+			weight: 4,
+			properties: {
+				rating: [1, 2, 3, 3, 4, 4, 4, 5, 5, 5],
+				feedback_type: ["accuracy", "relevance", "timeliness", "completeness", "tone"],
+				agent_type: pickAWinner(["meeting_assistant", "follow_up_writer", "crm_updater", "pre_call_researcher"]),
+			},
+		},
+		{
+			event: "dashboard viewed",
+			weight: 7,
+			properties: {
+				dashboard_type: ["pipeline_overview", "team_performance", "ai_agent_stats", "revenue_forecast", "activity_feed"],
+				time_spent_sec: weighNumRange(10, 300, 0.3),
+			},
+		},
+		{
+			event: "integration connected",
+			weight: 1,
+			properties: {
+				integration_name: pickAWinner(["hubspot", "salesforce", "slack", "google_calendar", "zoom", "linkedin", "gmail", "outlook"]),
+				setup_time_min: weighNumRange(2, 30, 0.5),
+			},
+		},
+		{
+			event: "search performed",
+			weight: 3,
+			properties: {
+				search_type: ["deals", "contacts", "meetings", "insights", "activities"],
+				results_count: weighNumRange(0, 50, 0.4),
+				clicked_result: [true, true, true, false],
+			},
+		},
+		{
+			event: "workflow created",
+			weight: 2,
+			properties: {
+				workflow_type: ["meeting_prep", "post_meeting", "deal_update", "weekly_digest", "pipeline_review"],
+				steps_count: weighNumRange(2, 8, 0.5),
+				trigger_type: ["time_based", "event_based", "manual"],
+			},
+		},
+		{
+			event: "contact enriched",
+			weight: 5,
+			properties: {
+				enrichment_source: ["linkedin", "company_website", "crm_history", "email_threads", "meeting_notes"],
+				fields_enriched: weighNumRange(2, 12, 0.4),
+				confidence_level: ["high", "high", "medium", "medium", "low"],
+			},
+		},
+		{
+			event: "deal created",
+			weight: 2,
+			properties: {
+				deal_source: ["inbound", "outbound", "referral", "upsell", "expansion"],
+				estimated_value: weighNumRange(5000, 250000, 0.2),
+				close_date_days: weighNumRange(14, 120, 0.3),
+			},
+		},
+		{
+			event: "coaching tip viewed",
+			weight: 3,
+			properties: {
+				tip_category: ["objection_handling", "discovery_questions", "closing_techniques", "rapport_building", "negotiation"],
+				tip_relevance_score: weighNumRange(40, 98, 0.4),
+				applied: [true, false, false, false],
+			},
+		},
+	],
+
+	// ──────────────────────────────────────────────
+	// FUNNELS (5)
+	// ──────────────────────────────────────────────
+	funnels: [
+		{
+			name: "Onboarding",
+			sequence: ["account created", "integration connected", "agent deployed", "meeting joined"],
+			isFirstFunnel: true,
+			conversionRate: 65,
+			timeToConvert: 48,
+			order: "sequential",
+			weight: 1,
+		},
+		{
+			name: "Meeting to Action",
+			sequence: ["meeting joined", "meeting summary generated", "follow-up drafted", "follow-up sent"],
+			conversionRate: 70,
+			timeToConvert: 4,
+			order: "sequential",
+			weight: 8,
+		},
+		{
+			name: "Pipeline Management",
+			sequence: ["pipeline alert received", "dashboard viewed", "deal stage changed"],
+			conversionRate: 45,
+			timeToConvert: 24,
+			order: "sequential",
+			weight: 5,
+		},
+		{
+			name: "AI Adoption",
+			sequence: ["agent deployed", "insight generated", "agent feedback"],
+			conversionRate: 55,
+			timeToConvert: 72,
+			order: "sequential",
+			weight: 4,
+		},
+		{
+			name: "Value Realization",
+			sequence: ["dashboard viewed", "workflow created", "agent deployed"],
+			conversionRate: 40,
+			timeToConvert: 96,
+			order: "sequential",
+			weight: 3,
+		},
+	],
+
+	// ──────────────────────────────────────────────
+	// SUPER PROPS (3) — on every event
+	// ──────────────────────────────────────────────
+	superProps: {
+		subscription_tier: pickAWinner(["starter", "growth", "growth", "enterprise"]),
+		team_role: pickAWinner(["sales_rep", "sales_rep", "sales_rep", "sales_manager", "rev_ops", "marketing"]),
+		ai_agent_version: ["v1.0", "v1.0", "v1.5", "v1.5", "v1.5", "v2.0", "v2.0", "v2.0"],
+	},
+
+	// ──────────────────────────────────────────────
+	// USER PROPS (5)
+	// ──────────────────────────────────────────────
+	userProps: {
+		company_size: pickAWinner(["SMB", "mid_market", "enterprise"]),
+		industry: pickAWinner(["SaaS", "fintech", "healthcare", "e_commerce", "consulting", "manufacturing"]),
+		deals_in_pipeline: weighNumRange(2, 50, 0.3),
+		onboarding_cohort: ["Q1_2025", "Q2_2025", "Q3_2025", "Q4_2025"],
+		integrations_connected: weighNumRange(1, 8, 0.4),
+	},
+
+	// ──────────────────────────────────────────────
+	// GROUPS (1 group key)
+	// ──────────────────────────────────────────────
+	groupKeys: [
+		["company_id", 500],
+	],
+
+	groupProps: {
+		company_id: {
+			name: () => chance.company(),
+			industry: pickAWinner(["SaaS", "fintech", "healthcare", "e_commerce", "consulting", "manufacturing"]),
+			segment: ["SMB", "mid_market", "enterprise"],
+			arr: weighNumRange(10000, 500000, 0.2),
+			csm: () => chance.pickone(["Sarah Chen", "Marcus Johnson", "Priya Patel", "Alex Rivera", "Jordan Kim"]),
+		},
+	},
+
+	// ──────────────────────────────────────────────
+	// HOOK FUNCTION (8 hooks)
+	// ──────────────────────────────────────────────
+	hook: function (record, type, meta) {
+		const NOW = dayjs();
+
+		// ════════════════════════════════════════════
+		// HOOK 1 — AI Model Quality Evolution (EVENT)
+		// Meeting summary quality improves after day 60
+		// ════════════════════════════════════════════
+		// HOOK 2 — Personalization Drives Outcomes (EVENT)
+		// Higher personalization → better email outcomes
+		// ════════════════════════════════════════════
+		// HOOK 3 — Weekend Autonomous Agents (EVENT)
+		// AI works autonomously on weekends
+		// ════════════════════════════════════════════
+		// HOOK 4 — Deal Velocity by Tier (EVENT)
+		// Enterprise deals move faster through pipeline
+		// ════════════════════════════════════════════
+		if (type === "event") {
+			const EVENT_TIME = dayjs(record.time);
+			const QUALITY_INFLECTION = NOW.subtract(60, "day");
+
+			// Hook 1: AI Model Quality Evolution
+			if (record.event === "meeting summary generated") {
+				if (EVENT_TIME.isBefore(QUALITY_INFLECTION)) {
+					record.quality_score = Math.round(decimal(50, 72) * 10) / 10;
+					record.ai_model = "v1_base";
+				} else {
+					const daysSinceInflection = Math.min(60, EVENT_TIME.diff(QUALITY_INFLECTION, "day"));
+					const improvement = (daysSinceInflection / 60) * 0.15;
+					record.quality_score = Math.min(98, Math.round(decimal(75, 95) * (1 + improvement) * 10) / 10);
+					record.ai_model = "v2_enhanced";
+				}
+			}
+
+			// Hook 2: Personalization Drives Outcomes
+			if (record.event === "follow-up sent") {
+				const persScore = record.personalization_score || 50;
+				if (persScore > 70) {
+					record.email_outcome = chance.pickone(["opened", "opened", "opened", "clicked", "clicked", "replied", "replied"]);
+					record.personalization_level = "high";
+				} else if (persScore > 50) {
+					record.email_outcome = chance.pickone(["pending", "opened", "opened", "clicked", "bounced"]);
+					record.personalization_level = "medium";
+				} else {
+					record.email_outcome = chance.pickone(["pending", "pending", "bounced", "bounced", "opened"]);
+					record.personalization_level = "low";
+				}
+			}
+
+			// Hook 3: Weekend Autonomous Agents
+			if (record.event === "crm auto-updated") {
+				const eventDay = EVENT_TIME.day(); // 0=Sun, 6=Sat
+				if (eventDay === 0 || eventDay === 6) {
+					record.update_source = "ai_autonomous";
+					record.fields_updated = Math.max(record.fields_updated || 1, integer(3, 8));
+				} else {
+					record.update_source = chance.pickone(["ai_assisted", "ai_assisted", "ai_assisted", "ai_autonomous"]);
+				}
+			}
+
+			// Hook 4: Deal Velocity by Tier
+			if (record.event === "deal stage changed") {
+				const tier = record.subscription_tier;
+				if (tier === "enterprise") {
+					record.days_in_previous_stage = Math.max(1, Math.round((record.days_in_previous_stage || 15) * decimal(0.5, 0.8)));
+					record.velocity_category = "fast";
+				} else if (tier === "starter") {
+					record.days_in_previous_stage = Math.round((record.days_in_previous_stage || 15) * decimal(1.3, 2.0));
+					record.velocity_category = "slow";
+				} else {
+					record.velocity_category = "normal";
+				}
+			}
+		}
+
+		// ════════════════════════════════════════════
+		// HOOK 5 — Power User + Integration Depth (USER)
+		// Tags users with engagement segments
+		// ════════════════════════════════════════════
+		if (type === "user") {
+			const intCount = record.integrations_connected || 1;
+			if (intCount >= 5) {
+				record.data_completeness_tier = "high";
+			} else if (intCount >= 3) {
+				record.data_completeness_tier = "medium";
+			} else {
+				record.data_completeness_tier = "low";
+			}
+
+			const tier = record.subscription_tier;
+			if (tier === "enterprise") {
+				record.user_segment = chance.pickone(["champion", "champion", "champion", "power_user", "standard"]);
+			} else if (tier === "growth") {
+				record.user_segment = chance.pickone(["champion", "power_user", "power_user", "standard", "standard"]);
+			} else {
+				record.user_segment = chance.pickone(["explorer", "explorer", "standard", "standard", "at_risk"]);
+			}
+		}
+
+		// ════════════════════════════════════════════
+		// HOOK 6 — Tier-Based Conversion (FUNNEL-PRE)
+		// Enterprise converts better, starter struggles
+		// ════════════════════════════════════════════
+		if (type === "funnel-pre") {
+			const tier = meta?.profile?.subscription_tier;
+			if (tier === "enterprise") {
+				record.conversionRate = Math.min(99, Math.round(record.conversionRate * decimal(1.2, 1.4)));
+			} else if (tier === "starter") {
+				record.conversionRate = Math.max(5, Math.round(record.conversionRate * decimal(0.6, 0.8)));
+			}
+		}
+
+		// ════════════════════════════════════════════
+		// HOOK 7 — Agent Adoption → Power Users (EVERYTHING)
+		// Early agent adoption predicts retention
+		// ════════════════════════════════════════════
+		// HOOK 8 — Negative Feedback → Disengagement (EVERYTHING)
+		// Repeated bad feedback predicts churn
+		// ════════════════════════════════════════════
+		if (type === "everything") {
+			if (!record.length) return record;
+			const userId = record[0].user_id || record[0].device_id;
+			if (!userId) return record;
+
+			// Hook 7: Agent Adoption → Power Users
+			const times = record.map(e => new Date(e.time).getTime()).filter(t => !isNaN(t));
+			if (times.length) {
+				const firstEventTime = dayjs(Math.min(...times));
+				const thirtyDayMark = firstEventTime.add(30, "day");
+
+				const earlyAgentDeploys = record.filter(e =>
+					e.event === "agent deployed" && dayjs(e.time).isBefore(thirtyDayMark)
+				).length;
+				const earlyMeetings = record.filter(e =>
+					e.event === "meeting joined" && dayjs(e.time).isBefore(thirtyDayMark)
+				).length;
+
+				if (earlyAgentDeploys >= 3 && earlyMeetings >= 5) {
+					record.push({
+						event: "power user milestone",
+						time: thirtyDayMark.add(integer(1, 48), "hour").toISOString(),
+						user_id: userId,
+						insert_id: uid(12),
+						milestone_type: "early_adopter",
+						agents_deployed: earlyAgentDeploys,
+						meetings_recorded: earlyMeetings,
+					});
+				} else if (earlyAgentDeploys < 1) {
+					record.push({
+						event: "churn risk flagged",
+						time: thirtyDayMark.add(integer(1, 72), "hour").toISOString(),
+						user_id: userId,
+						insert_id: uid(12),
+						risk_reason: "low_agent_adoption",
+						agents_deployed: earlyAgentDeploys,
+					});
+				}
+			}
+
+			// Hook 8: Negative Feedback → Disengagement
+			const negativeFeedbacks = record.filter(e =>
+				e.event === "agent feedback" && (e.rating === 1 || e.rating === 2)
+			).length;
+
+			if (negativeFeedbacks >= 3) {
+				const sortedByTime = [...record].sort((a, b) => new Date(a.time) - new Date(b.time));
+				const midpoint = Math.floor(sortedByTime.length * 0.6);
+				const cutoffTime = sortedByTime[midpoint]?.time;
+
+				if (cutoffTime) {
+					const filtered = record.filter(e => {
+						if (new Date(e.time) > new Date(cutoffTime)) {
+							return chance.bool({ likelihood: 40 });
+						}
+						return true;
+					});
+					record.length = 0;
+					record.push(...filtered);
+				}
+			}
+		}
+
+		return record;
+	},
+};
+
+export default config;
+
+/**
+ * ═══════════════════════════════════════════════════════════════════
+ * CLINCH AGI — DATASET DOCUMENTATION
+ * ═══════════════════════════════════════════════════════════════════
+ *
+ * OVERVIEW
+ * --------
+ * This dungeon generates ~120,000 events across 1,000 users over 120 days,
+ * simulating Clinch AGI — an AI-native workflow automation platform for
+ * sales and go-to-market teams. The data models core product behaviors:
+ * AI meeting assistants, automated CRM updates, pipeline intelligence,
+ * personalized outreach, and AI coaching.
+ *
+ * 200 companies are generated as group analytics entities.
+ *
+ * HOOKS — 8 DELIBERATELY ARCHITECTED ANALYTICS PATTERNS
+ * ─────────────────────────────────────────────────────
+ *
+ * ┌──────┬────────────────────────────────┬────────────┬─────────────────────────────────────────────────┐
+ * │ Hook │ Name                           │ Type       │ How to Find It in Mixpanel                      │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  1   │ AI Model Quality Evolution     │ event      │ Insights → Line chart: avg(quality_score) on    │
+ * │      │                                │            │ "meeting summary generated" over time.           │
+ * │      │                                │            │ Clear inflection ~60 days ago. Breakdown by      │
+ * │      │                                │            │ ai_model shows v1_base vs v2_enhanced.           │
+ * │      │ Real-world: Model improvement  │            │                                                 │
+ * │      │ after training data refinement │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  2   │ Personalization Drives Results │ event      │ Insights → Bar chart: "follow-up sent" broken   │
+ * │      │                                │            │ down by email_outcome, filtered by               │
+ * │      │                                │            │ personalization_level. High personalization has   │
+ * │      │                                │            │ 3x more replies vs low.                          │
+ * │      │ Real-world: AI effort → ROI    │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  3   │ Weekend Autonomous Agents      │ event      │ Insights → "crm auto-updated" broken down by    │
+ * │      │                                │            │ update_source, viewed by day of week. Weekends   │
+ * │      │                                │            │ are nearly 100% ai_autonomous with higher        │
+ * │      │                                │            │ fields_updated. AI works when humans rest.       │
+ * │      │ Real-world: 24/7 AI coverage   │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  4   │ Deal Velocity by Tier          │ event      │ Insights → avg(days_in_previous_stage) on       │
+ * │      │                                │            │ "deal stage changed", broken down by             │
+ * │      │                                │            │ subscription_tier. Enterprise is 2x faster.      │
+ * │      │                                │            │ Also see velocity_category breakdown.             │
+ * │      │ Real-world: Tier-driven ROI    │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  5   │ Power User + Integration Depth │ user       │ Users → Breakdown by user_segment shows          │
+ * │      │                                │            │ champion/power_user/standard/explorer/at_risk.   │
+ * │      │                                │            │ Cross with data_completeness_tier to see          │
+ * │      │                                │            │ integration depth correlates with user quality.   │
+ * │      │ Real-world: PQL scoring model  │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  6   │ Tier-Based Conversion          │ funnel-pre │ Funnels → Any funnel broken down by              │
+ * │      │                                │            │ subscription_tier. Enterprise converts 1.3x      │
+ * │      │                                │            │ better; starter converts 0.7x worse.             │
+ * │      │ Real-world: Segment activation │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  7   │ Agent Adoption → Power Users   │ everything │ Retention → Cohort by "power user milestone"    │
+ * │      │                                │            │ vs "churn risk flagged". Milestone users retain   │
+ * │      │                                │            │ at 3x the rate. Filter by agents_deployed to     │
+ * │      │                                │            │ see the adoption threshold.                       │
+ * │      │ Real-world: Aha moment         │            │                                                 │
+ * ├──────┼────────────────────────────────┼────────────┼─────────────────────────────────────────────────┤
+ * │  8   │ Negative Feedback → Churn      │ everything │ Retention → Segment by users who did             │
+ * │      │                                │            │ "agent feedback" with rating 1 or 2, >=3 times.  │
+ * │      │                                │            │ These users show steep engagement dropoff in      │
+ * │      │                                │            │ second half of their lifecycle.                   │
+ * │      │ Real-world: Churn prediction   │            │                                                 │
+ * └──────┴────────────────────────────────┴────────────┴─────────────────────────────────────────────────┘
+ *
+ * EXPECTED METRICS SUMMARY
+ * ────────────────────────
+ * │ Metric                         │ Expected Range          │
+ * │ Total events                   │ ~120,000                │
+ * │ Unique users                   │ ~1,000                  │
+ * │ Companies (groups)             │ 200                     │
+ * │ Avg quality_score (early)      │ 50-72                   │
+ * │ Avg quality_score (recent)     │ 78-95                   │
+ * │ Enterprise deal velocity       │ 2-24 days avg           │
+ * │ Starter deal velocity          │ 15-120 days avg         │
+ * │ Onboarding funnel conversion   │ ~65% (enterprise ~85%)  │
+ * │ Power user milestone rate      │ ~15-25% of users        │
+ * │ Churn risk flagged rate        │ ~30-40% of users        │
+ *
+ * CROSS-HOOK ANALYSIS IDEAS
+ * ─────────────────────────
+ * - Do power users (Hook 7) also have high data_completeness_tier (Hook 5)?
+ * - Does AI model improvement (Hook 1) correlate with better email outcomes (Hook 2)?
+ * - Do enterprise tier users (Hook 4, 6) generate more positive agent feedback (Hook 8)?
+ * - Is weekend autonomous behavior (Hook 3) more common for growth/enterprise tiers?
+ */

package/dungeons/complex.js

@@ -8,12 +8,12 @@
 
 import Chance from 'chance';
 const chance = new Chance();
-import { weighNumRange, date, integer } from '../lib/utils/utils.js';
+import { weighNumRange, date, integer } from "../lib/utils/utils.js";
 import * as u from 'ak-tools';
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	token: "",
+	token: process.env.MASTER_PROJECT_TOKEN || "",
 	seed: "quite complexus",
 	numDays: 30, //how many days worth of data
 	numEvents: 100_000, //how many events
@@ -33,6 +33,7 @@ const config = {
 	hasAdSpend: true,
 
 	hasAvatar: true,
+	makeChart: false,
 
 	batchSize: 500_000,
 	concurrency: 10,