# Customer Churn Root-Cause Diagnosis ## Q2 Churned Cohort (n=45) — B2B Project Management SaaS, $49/seat/month, Mid-Market (20–200 employees) --- ### CARD 1 — EXECUTIVE SUMMARY **Leading hypothesis (if any):** Onboarding/product-complexity friction — most visibly manifest in abandonment of the core reporting feature — is driving disengagement and eventual churn, tied most strongly to the 70% reporting-feature usage collapse (DOCUMENTED, sized against a stable comparison cohort). **Evidence-quality warning:** Three signals are DOCUMENTED and quantified, each from a distinct category (Usage Decline, Support Friction, Sentiment). Structural convergence is strong. However, the specific thread tying all three into *one* driver — that the tickets and the "too complicated to set up" comments are actually *about* the reporting feature — is INFERRED, not directly stated in the evidence. No Commercial or Relationship Signal data was supplied, and all figures are cohort-level averages, which can mask account-level heterogeneity across the 45 accounts. --- ### CARD 2 — SIGNAL INVENTORY BY CATEGORY | Category | Signal | Status | Evidence Locator | |---|---|---|---| | Usage Decline Signals | Core reporting-feature usage down 70% in final month pre-cancellation (cohort avg) | DOCUMENTED SIGNAL | Cohort usage metric, final month | | Usage Decline Signals | Comparison cohort shows stable reporting-feature usage in same window | DOCUMENTED SIGNAL (context) | Comparison cohort data | | Support Friction Signals | Avg 3.0 support tickets/account in final 60 days vs. 0.8 company-wide avg | DOCUMENTED SIGNAL | Cohort ticket counts, final 60 days | | Support Friction Signals | Comparison (retained) cohort averaged 1.1 tickets in same window | DOCUMENTED SIGNAL (context) | Comparison cohort data | | Sentiment Signals | NPS comments repeatedly cite "too complicated to set up" | DOCUMENTED SIGNAL | NPS comment text (exact count/frequency not given) | | Commercial Signals | None supplied | UNKNOWN | — | | Relationship Signals | None supplied | UNKNOWN | — | --- ### CARD 3 — SIGNAL-BY-SIGNAL REGISTER | Signal | Category | Evidence Locator | Status | Churn-Relevance | |---|---|---|---|---| | Elevated ticket volume (3.0 vs. 0.8 company avg; 3.0 vs. 1.1 retained-cohort avg) | Support Friction | Cohort ticket data, final 60 days | DOCUMENTED | HIGH — ~3.75x company baseline, ~2.7x retained-cohort baseline | | "Too complicated to set up" NPS theme | Sentiment | NPS comment text | DOCUMENTED | HIGH — direct, unprompted usability complaint | | Reporting-feature usage down 70% pre-cancellation | Usage Decline | Cohort usage metric | DOCUMENTED | HIGH — severe, and diverges sharply from stable retained cohort | | Ticket topic/content not categorized | Support Friction (gap) | — | UNKNOWN | Limits confidence that tickets relate to "complexity" theme | | Frequency of "too complicated" comment across the 45 accounts unquantified | Sentiment (gap) | — | UNKNOWN | Limits confidence in how cohort-wide the sentiment actually is | | No billing/renewal friction data | Commercial | — | UNKNOWN / NOT PROVIDED | Cannot assess as contributing driver | | No champion/stakeholder turnover data | Relationship | — | UNKNOWN / NOT PROVIDED | Cannot assess as contributing driver | --- ### CARD 4 — DRIVER DIAGNOSTIC **D1 — Onboarding/Setup-Complexity-Driven Disengagement** | Factor | Score | Status | Rationale | |---|---|---|---| | Signal Strength | HIGH | DOCUMENTED | Anomalies far exceed comparison cohort: ~2.7–3.75x ticket volume; usage collapse vs. stable retained-cohort usage | | Signal Category Breadth | 3 categories (Support Friction, Sentiment, Usage Decline) | DOCUMENTED (breadth); cross-signal topical link is INFERRED | Each signal is individually documented; their shared "complexity" identity is an inference, not a stated fact | | Recency | HIGH | DOCUMENTED | All signals concentrated in the final 60 days / final month pre-cancellation | | Signal Convergence Index (OPERATIONAL EXTENSION — NOT A CAUSATION MEASURE) | HIGH | — | 3 independent categories align in direction and timing, satisfying the breadth requirement — but this reflects co-occurrence, not confirmed shared causation | **Core Driver Reading:** The three signals point the same direction, at the same time, at unusually large magnitudes relative to a matched retained cohort. That pattern is the strongest evidenced story available. The gap is that nothing in the evidence confirms the tickets were *about* setup/complexity, or that the "too complicated" comments referenced the *reporting feature* specifically — the unifying narrative is plausible and well-timed, not confirmed. **D2 — Reporting-Feature Non-Adoption (narrower framing, independent of the complexity theme)** | Factor | Score | Status | Rationale | |---|---|---|---| | Signal Strength | HIGH | DOCUMENTED | 70% usage drop is large and specific to one feature | | Signal Category Breadth | 1 category (Usage Decline) if treated independently of D1's sentiment signal | DOCUMENTED | Standing alone, this doesn't require the "complexity" inference | | Recency | HIGH | DOCUMENTED | Occurs in the final month pre-cancellation | | Signal Convergence Index | MODERATE | — | Per Rule 4, a single-category signal is capped at MODERATE regardless of magnitude, unless linked to D1 | **Core Driver Reading:** Taken on its own, without assuming the setup complaints are about reporting specifically, this is a well-evidenced but narrower driver: customers stopped using the feature the product is presumably built around, for a reason the data doesn't specify. **D3 — General Support-Experience Dissatisfaction (independent of stated complexity theme)** | Factor | Score | Status | Rationale | |---|---|---|---| | Signal Strength | MODERATE | DOCUMENTED (volume); UNVERIFIED (dissatisfaction narrative) | Ticket volume is documented; whether tickets reflect *dissatisfaction* vs. simple help-seeking is not established (no resolution/CSAT data) | | Signal Category Breadth | 1 category (Support Friction) | DOCUMENTED | — | | Recency | HIGH | DOCUMENTED | Final 60 days | | Signal Convergence Index | LOW–MODERATE | — | Single category; causal story beyond raw volume is an UNVERIFIED HYPOTHESIS | --- ### CARD 5 — RANKED LIKELY DRIVERS 1. **D1 — Onboarding/Setup-Complexity-Driven Disengagement.** Confidence: HIGH structurally, tempered by an inferred cross-signal link. Strongest signal: three categories (tickets, sentiment, usage) all anomalous, all concentrated in the same pre-churn window. 2. **D2 — Reporting-Feature Non-Adoption.** Confidence: MODERATE. Strongest signal: 70% usage drop against a comparison cohort holding steady. 3. **D3 — General Support-Experience Dissatisfaction.** Confidence: LOW–MODERATE. Strongest signal: ~3.75x ticket volume vs. company average, but causal story unconfirmed. --- ### CARD 6 — CONTRADICTING/COMPLICATING EVIDENCE - **Ticket content is unknown.** Nothing confirms the elevated tickets are about setup, configuration, or the reporting feature — they could reflect billing questions, bugs, or feature requests unrelated to "complexity." - **Comment frequency is unquantified.** "Repeatedly mentions" doesn't say whether this is 5 of 45 accounts or 30 of 45 — the sentiment signal's cohort-wide reach is uncertain. - **Averages can mask heterogeneity.** All three headline figures are cohort averages. It's possible a subset of accounts drives most of the anomaly while others churned for unrelated reasons — the evidence doesn't establish that most or all 45 accounts show all three signals together. - **No usage-vs-sentiment divergence found.** Per Rule 9, this was explicitly checked: usage (down) and sentiment (negative) point the same direction here, not opposite ones, so there is no directional contradiction to flag on that axis. - **Two full signal categories are absent.** No Commercial Signals (billing, renewal friction, pricing) and no Relationship Signals (champion turnover) were supplied — a driver from either category can't be ruled in or out. --- ### CARD 7 — COHORT COMPARISON NOTE The comparison cohort sizes, but does not explain, the anomaly. Retained accounts of similar size averaged 1.1 tickets in the same 60-day window (churned cohort: 3.0, ~2.7x higher) and held stable reporting-feature usage (churned cohort: −70%). This confirms the churned cohort's signals are unusual relative to a matched baseline — it does not establish that setup complexity, specifically, caused either the ticket spike or the usage drop. --- ### CARD 8 — SAVE-PLAY OPTIONS | Driver | Intervention | Owner | Confirmation Signal | |---|---|---|---| | D1 | Onboarding audit + revamped reporting-feature setup flow (guided walkthrough, templates) targeted at accounts showing early ticket/complexity patterns | Customer Success / Onboarding | Drop in setup-related ticket volume; stabilized reporting-feature weekly active usage | | D2 | Dedicated reporting-feature re-engagement: live training session or in-app nudges for accounts with declining feature usage | Product / CS | Reporting-feature usage returns toward baseline within a defined window | | D3 | Ticket-topic tagging and CSAT tracking to determine whether the volume spike reflects genuine dissatisfaction or simple help-seeking | Support Ops | Ticket category breakdown available; CSAT trend on high-volume accounts | --- ### CARD 9 — MISSING DATA THAT WOULD SHARPEN THE DIAGNOSIS - Ticket topic/category tagging for the 45 churned accounts - Distribution (not just averages) of ticket counts, usage decline %, and comment sentiment across the 45 accounts, to check for subgroup heterogeneity - Count/proportion of the 45 accounts whose comments specifically referenced "too complicated to set up" - Whether "too complicated to set up" referred to the reporting feature or to general onboarding/workspace setup - Commercial Signals: pricing changes, renewal friction, plan downgrades prior to churn - Relationship Signals: champion or admin turnover before cancellation - Sequencing: did the ticket spike precede or follow the reporting-feature usage drop - Whether churned accounts ever adopted the reporting feature at onboarding, or abandoned it after initial use - Cancellation-reason or exit-survey data, if collected --- ### CARD 10 — METHOD NOTE This diagnosis is a structured aid for prioritizing intervention, not a certainty. It applies the five-category Churn Signal Taxonomy (Usage Decline, Support Friction, Sentiment, Commercial, Relationship) plus the Signal Convergence Index — an operational, non-causal measure of how many independent categories point the same direction, used here only because signals from three categories (Support Friction, Sentiment, Usage Decline) converged in timing and direction. The comparison cohort was used solely to size the anomaly (ticket volume ~2.7x higher, usage decline vs. stable baseline), never as proof of what caused it. Two signal categories (Commercial, Relationship) had no supporting evidence and are flagged as gaps rather than assumed absent of risk. Direct customer conversation with a sample of these 45 accounts — or ticket-topic tagging — remains the most reliable way to confirm whether "complexity" is genuinely the shared thread across these three signals. --- **VERDICT: CLEAR DRIVER IDENTIFIED — Onboarding/product-complexity friction, most visibly manifest in reporting-feature abandonment, is the leading driver, with HIGH signal convergence across Usage Decline, Support Friction, and Sentiment categories; the specific link tying ticket volume and "too complicated to set up" comments to the reporting feature itself remains INFERRED rather than directly documented.**