- Project type
- Portfolio project
- Dataset
- Synthetic dataset
- Dataset provenance
- A synthetic SaaS dataset (six tables, 86,402 rows, 2022 to 2024) generated with a reproducible Python script.
- Client status
- Demonstration analysis; does not represent a PA Data Analytics client.
View the data behind this chart
| Health category | Accounts | MRR |
|---|---|---|
| Healthy | 1,315 | $288,465 |
| Medium Risk | 332 | $92,598 |
| High Risk | 131 | $21,489 |
| Months since signup | Average retention |
|---|---|
| 1 | 91.6% |
| 3 | 81.3% |
| 6 | 72.0% |
| 12 | 60.2% |
01Business question
A subscription business can usually see overall revenue growth, but not whether that revenue is healthy: where monthly recurring revenue (MRR) is leaking, which customers are genuinely engaged, and which accounts show warning signs before they cancel.
02Dataset & context
Dataset: Synthetic / Demonstration Dataset. This portfolio analysis uses a synthetic SaaS dataset generated with a reproducible Python script: six tables (customers, marketing, subscriptions, payments, product usage and support tickets) with 86,402 rows covering 2022 to 2024. It does not represent a PA Data Analytics client.
03Approach
Break revenue into its movements (new, expansion, contraction, churn and reactivation), track cohort retention, compare acquisition channels on lifetime value, and then look for usage and support signals that appear before cancellation.
04Methodology
- MRR trend and MRR waterfall built from the subscription history.
- Monthly cohort retention and plan-level revenue analysis.
- LTV:CAC by acquisition channel.
- Comparison of product usage and support satisfaction (CSAT) for customers before cancellation versus other customers.
- A rule-based customer health score (0–100) weighting usage trend (40 points), CSAT (30), support ticket volume (20) and recent activity (10). This is a scoring framework, not a trained machine-learning model.
- SQL views and Looker Studio-ready exports.
05Modelling & analysis
Dataset findings:
- Contraction MRR from downgrades ($131,788) exceeded churned MRR from cancellations ($115,124), so plan downgrades cost more than full cancellations.
- Enterprise plans accounted for 64.8% of active MRR.
- Average cohort retention was 91.6% at month 1, 81.3% at month 3, 72.0% at month 6 and 60.2% at month 12.
- Weekly product sessions fell from an average of 95.2 to 20.3 (a 79% drop) in the weeks before cancellation.
- Customers about to cancel had an average CSAT of 2.44/5, compared with 3.74/5 for other customers.
06Results within the dataset
The health score flagged 131 high-risk accounts ($21,489 MRR) and 332 medium-risk accounts ($92,598 MRR): $114,087 of MRR across 463 accounts showing warning signals in the dataset. The score's weights are analyst-set assumptions, and the figures describe the synthetic data, not a real company's revenue.
07What this analysis demonstrates
How revenue movements, retention, acquisition quality and behavioural signals can be combined to show where subscription revenue is at risk. It also shows that falling usage and satisfaction can serve as early warning signs well before a cancellation is recorded.
08How a business could use this analysis
- Give customer success teams a ranked list of at-risk accounts instead of contacting every customer equally.
- Monitor weekly usage and CSAT as leading indicators of churn.
- Evaluate acquisition channels on LTV:CAC, not acquisition cost alone.
- Validate the health score against actual cancellations over time, or replace it with a trained model once enough history exists.
09Related service
This project demonstrates methods used in our customer intelligence services.
10Apply this to your data
If you are facing a similar question, we can look at what your data can support and what a useful answer would look like.