Back to projects
Case StudyCustomer Intelligence & Retention

Unified Customer Intelligence & Retention Platform

As a product leader in an enterprise data management company, I designed and led the rollout of a unified customer intelligence layer that consolidated Jira, HubSpot, usage signals, and sentiment data into a single view of customer health. The platform turned fragmented operational data into proactive retention and roadmap signals.

Customer IntelligenceChurn & RetentionData ProductsAI & Automation
Outcome Snapshot

Retention Lift (Monitored Cohort)

+18%

Time to Prepare QBR Insights

-95%

Accounts With Live Health Scores

90%

Unified view across Jira, HubSpot, and product usage.

Roadmap Items Tied to Signals

85%

Product decisions backed by customer evidence.

Role & Scope

Product lead for a cross-functional initiative spanning Customer Success, Support, Product, Data Engineering, and Sales.

Accountability across discovery, product strategy, requirements, prioritisation, stakeholder alignment, and launch.

Tech Stack
Next.jsTypeScriptPythonSnowflake / PostgresJiraHubSpotNLP / LLM (OpenAI API)
North Star Metric

Signal-to-Action Time

Average time from a customer signal (incident, sentiment drop, usage anomaly) to an internal action or follow-up.

Unified Customer Intelligence Dashboard (Mock)

This mock dashboard illustrates how Customer Success, Product, and Sales teams consume the unified signals: health scores, churn risk, SLA performance, sentiment, and roadmap drivers.

Accounts with live health score

90%

+65 pts

High-risk accounts (flagged)

12

-6

Avg. QBR prep time

28 min

-9.3 hrs

Roadmap items tied to signals

85%

+60 pts
Health Score Distribution

Accounts grouped into health bands, updated every 15 minutes from Jira, HubSpot, and usage logs.

Healthy (80–100)58%
Watchlist (60–79)27%
At-risk (40–59)10%
Critical (<40)5%
Churn Risk Radar

Top drivers of churn risk across the portfolio.

SLA Breaches (P1/P2)High risk
Negative Sentiment (Email/Tickets)High risk
Feature Gaps in Critical WorkflowsMedium risk
Low Executive SponsorshipMedium risk
Low Feature Adoption (New Releases)Medium risk
Insight Feed → Product Backlog
Feature gap15 accounts

Data lineage visualisation missing in regulatory reporting workflow.

High ARR coverage • aligns to compliance-led roadmap theme.

Reliability9 incidents • 3 clients

Intermittent latency spikes on overnight batch load impacting start-of-day.

Linked to SLA & MTTR KPIs • routed to platform team.

UX friction23 tickets

Confusing permissions model causing onboarding delays for new users.

Cross-functional CX theme • candidate for design sprint.

PRD Snapshot

Problem

Customer, incident, and product usage data were fragmented across Jira, HubSpot, email, and spreadsheets, making it hard to detect churn risk early and tie product decisions to clear evidence.

Goal & Non-Goals

  • Goal: Build a unified, near-real-time health and intelligence layer across all key accounts.
  • Goal: Reduce manual QBR/data prep while increasing the quality of insights used for roadmap and renewals.
  • Non-goal (v1): Full ML-based churn prediction. Start with rules-based scoring and an ML-ready schema.

Success Metrics

  • 80%+ of managed accounts have a live health score within 3 months.
  • Reduce QBR insight preparation time by >70% versus baseline.
  • 70%+ of roadmap items for the next two quarters are directly linked to customer intelligence signals.
  • Improve monitored cohort retention by >10% YoY.

Key Constraints & Trade-offs

  • Prioritised Jira + HubSpot integrations before usage logs for faster time-to-value.
  • Adopted off-the-shelf NLP for sentiment, with the option to swap/extend in future releases.
  • Scoped out real-time in-app hints; focused on internal dashboards + alerts for CS and Product teams.
System & Feedback Flow (Diagram)

High-level system view connecting data sources, intelligence layer, and end-users.

[Jira]   [HubSpot]   [Email]   [Usage Logs]
   \        |          |           /
    \       |          |          /
    [ETL & Event Pipeline (Python/AWS)]
              |
      [Customer Intelligence Layer]
      (Scoring + Sentiment + Signals)
              |
   +----------+------------+------------------+
   |                       |                  |
[CS Dashboard]      [Product Insights]    [Sales / AM View]
   |                       |                  |
   +--------- Feedback Loop into Roadmap & Renewals -----+

The key product decision: treat this as a platform capability that feeds multiple teams, not a single-team dashboard project.

Product Management Depth

This case study demonstrates my ability to design and deliver a complex data product that sits at the intersection of Customer Success, Product, and Revenue. It required systems thinking, strong stakeholder alignment, and a clear definition of constraints and trade-offs.

  • Applied PM frameworks (JTBD, RICE, opportunity solution tree, service blueprints) to shape scope.
  • Worked with data engineers, CS, sales, and product to define schemas, scoring logic, and UX for internal tools.
  • Led an iterative rollout: rules-based scoring first, then ML-ready architecture and AI-driven insights.

In interviews, I use this case study to show how I think about data products, retention, and the practical integration of AI into existing workflows.