Jenny Park Back to work

Product Analytics B2B SaaS

Product-Led SaaS Lifecycle & Retention Analytics

Mapped the full B2B SaaS customer lifecycle from first signup to churn risk — surfacing where accounts drop off, whether guided onboarding actually moves the needle, and which accounts need a customer success call before they churn.

Live Dashboard
Focus SaaS Lifecycle & Retention
Role Product & Growth Analyst
Methods Lifecycle Funnel · A/B Experiment · Cohort Retention · Churn Risk · NRR
Tools Python · Streamlit · Plotly · pytest
+18.6 pp Activation lift from guided onboarding — p < 0.001, statistically significant
$96K MRR exposed in medium-risk accounts — identifiable before any churn has occurred
93.5% 60-day retention for eligible paid accounts across 2,500 organizations

The Problem

Signup Numbers Looked Fine. The Lifecycle Told a Different Story.

Activation Gap: 35% of accounts never created a workspace after signing up — and that drop-off set a ceiling on everything downstream. Among those that did create a workspace, 68% went on to activate within 7 days. But that meant only 44% of all signups activated overall. No consistent definition of "activated" existed — and no way to connect early behavior to downstream paid conversion or churn.

Experiment Running Blind: A guided onboarding flow was live in production, but no analytic layer connected it to downstream outcomes. The team knew activation had improved — not whether that improvement carried into paid conversion, retention, or whether it worked the same way for all segments.

CS Outreach Based on Guesswork: Customer success was prioritizing accounts by size, not behavior. 1,112 medium-risk accounts carrying $96K in active MRR were showing declining engagement signals — with no system to surface them before they churned.

The data lived in five disconnected tables. Product behavior, subscription records, and experiment assignments had never been joined into a single lifecycle model. Every team saw their own slice. Nobody could see where accounts were leaking, or why.

Lifecycle Funnel

The Biggest Drop Happens Before Onboarding Even Begins

Joining five raw tables into a shared account-level model made the full lifecycle visible for the first time. The funnel reveals that the largest single leakage point — losing 35% of signups — happens at workspace creation, before any onboarding experiment or key product action has a chance to intervene.

Account Lifecycle — 2,500 Organizations
Signups 2,500  ·  100%
Workspace Created 35% lost here 1,615  ·  65%
Activated  (workspace + key action within 7 days) 1,106  ·  44%
Trial Started 738  ·  30%
Paid Customers 766  ·  31%
Retained 60-Day ~716  ·  93.5% of eligible
44% Overall 7-day activation rate
68% Of workspace creators go on to activate — the problem is before onboarding
93.5% 60-day retention once accounts reach paid
Key Insight
Among accounts that created a workspace, 68% went on to activate — a strong conversion rate. The problem is that 35% of signups never reach even that first milestone. Onboarding improvements help accounts that get through the door; they don't solve the drop before it.

A/B Experiment Readout

+18.6 pp Activation Lift. Paid Conversion: Flat.

The guided onboarding experiment (onboarding_guided_v1) compared the existing flow against a guided workspace setup with explicit prompts for the first key product action. 2,500 organizations were randomized at the account level.

Activation Rate — Control vs. Variant
p < 0.001 · Statistically significant
Primary Metric — 7-Day Activation Rate
Control
34.7%
1,212 accts
Variant
53.3%
+18.6 pp ↑
Paid Conversion Rate
Control 38.8%
Variant 38.7%
−0.1 pp — effectively unchanged
60-Day Retention
Control 94.3%
Variant 92.8%
−1.5 pp — within noise, monitor ongoing
The Tension
Activation improved by 18.6 pp. Paid conversion did not move. Activation lift that doesn't translate into monetization may reflect surface engagement — users completing onboarding steps without reaching genuine product value. Paid conversion needs to be a co-primary metric before this experiment is declared a success.

Segment-Level Treatment Effect

The lift was not uniform. SMB and Mid-Market showed large responses; Enterprise did not — likely because enterprise onboarding depends more on implementation support than in-product prompts.

SMB
+18 pp
activation lift
Control 33% → Variant 51%
✓ Roll out now
Mid-Market
+19 pp
activation lift
Control 34% → Variant 53%
✓ Roll out now
Enterprise
+7 pp
activation lift
Control 36% → Variant 43%
⚠ Separate path — monitor first

Churn Risk & CS Prioritization

$96K in MRR at Risk — Before Any Churn Has Occurred

Instead of a black-box propensity model, risk scoring uses four transparent signals any CS or product team can immediately verify: subscription status, 7-day activation, 60-day retention eligibility, and days since last product event. Each signal has an explicit point value — the score is auditable, not opaque.

Account distribution by risk segment
24.8%
44.5%
30.7%
High — 620 accounts
Medium — 1,112 accounts · $96K MRR
Low — 768 accounts
High Risk
620
24.8% of accounts. Churned or never converted. $0 MRR remaining — but represent the outcome pool for understanding what preventable churn looks like.
Medium Risk · Act Now
$96,627
1,112 accounts (44.5%). Active MRR with declining engagement. The highest-priority CS intervention target — revenue is still there, but signals are deteriorating.
Low Risk
768
30.7% of accounts. Active and engaged. Expansion opportunity — no immediate retention intervention required.
Key Insight
Medium-risk accounts are the operational priority — not because the risk score is highest, but because the MRR is still there and recoverable. High-risk accounts have already churned. Medium-risk accounts haven't — yet.

Recommendations

Four Decisions, Four Teams

1
Product
Roll out guided onboarding to SMB and Mid-Market now. Hold Enterprise.
The +18.6 pp activation lift is real and significant for SMB and Mid-Market. Enterprise showed only +7 pp — too small and too uncertain to justify the same flow. Enterprise onboarding needs implementation support, not in-product prompts. Separate the paths.
Success metric: 7-day activation rate by segment
2
Product
Move the teammate invite prompt earlier — before workspace setup completes.
Accounts that invite teammates show measurably higher 60-day retention than those that don't. This is currently a downstream event. Making it part of the activation flow turns a retention signal into a retention driver.
Success metric: Invite rate in first 7 days, 60-day retention delta
3
Analytics
Add paid conversion as a co-primary metric before declaring the experiment a success.
Activation is up 18.6 pp. Paid conversion is flat (−0.1 pp). Using activation alone as the success criterion would over-credit the experiment. Until conversion confirms the lift, rollout decisions should be segmented — not universal.
Success metric: Trial-to-paid conversion rate by variant and segment
4
Customer Success
Run a scaled re-engagement sequence for 1,112 medium-risk accounts.
$96K in active MRR is showing declining engagement. A targeted sequence — use-case reminder, training offer, and adoption review for highest-MRR accounts — can arrest churn before it happens. Measure 60-day survival rate as the programme outcome.
Success metric: 60-day survival rate in medium-risk cohort

Tech Stack

Tools & Methods

Python 3.12 pandas numpy Streamlit Plotly pytest Lifecycle Funnel A/B Experiment Analysis Two-Proportion Z-Test Cohort Retention NRR Modeling Churn Risk Scoring HTE Segmentation Event-Based Product Analytics Deterministic Synthetic Data

Source Code

Code & Documentation

Full Streamlit dashboard, data generation scripts, metric definitions, and test suite available on GitHub. The README includes a quickstart to run the full analytics system locally.

Live Dashboard View on GitHub

Interested in working together?

Open to Data Analyst, Product Analyst, Growth Analyst, and Revenue Analytics roles in the Bay Area or remote.

Get in touch See more work