I often get called in when a subscription business looks healthy on top-line MRR (monthly recurring revenue) but struggles to make predictable profits. The patterns are almost always the same: we’ve focused on acquisition, churn creeps up, and operational costs quietly expand. Mapping unit economics is the simplest way to turn intuition into a clear diagnosis — and it’s how I found a 20% margin leak in a recent client in under a week.
Why unit economics matter for subscription products
Unit economics tell you, at the level of a single customer (or cohort), whether your model is profitable and sustainable. For subscription businesses, that usually means calculating contribution margin per customer over a relevant time horizon. If you can’t answer “how much do I earn from one user after direct variable costs?” you're running blind.
When I map unit economics I want to know three things:
Core metrics and formulas I use
Here are the minimum metrics and formulas I put into a one-page model. Put them in a spreadsheet and make the cells editable so you can run scenarios quickly.
Tip: for subscription products with upgrades, downgrades or seasonal behavior, build a cohort model that tracks ARPU and churn over time rather than assuming constant numbers.
How I found a 20% margin leak — a short case study
Client: a B2B SaaS selling analytics at £49/month and £149/month tiers. They had ~1,200 customers and steady growth in MRR. The business felt profitable — until I mapped unit economics.
Initial inputs we pulled from billing and finance:
Gross margin per month = £84 - (£2.80 + £4 + £15) = £62.20 → ~74% gross margin. Sounds good. LTV = £62.20 * 25 = £1,555.
But then we layered in two often-missed costs and they changed the picture fast:
When I converted refunds and failed payments into per-user variable costs (refunds ≈ £2.52/user, failed payment losses ≈ £5/user), gross margin per month became £62.20 - £2.52 - £5 = £54.68 → ~65%. Over lifetime, LTV dropped from £1,555 to £1,367 — a 12% LTV decline. Combine that with a hidden onboarding cost: third-party integration credits and a high-touch onboarding team allocated at £30/user for the first 3 months but amortized incorrectly across all months instead of just onboarding months. Properly allocated, that pushed effective CAC and reduced margin further.
All together the margin leak amounted to roughly 18–22% of expected lifetime margin depending on cohort — in plain terms: what their CFO thought was a 40% contribution margin was actually closer to 20–25% after these variable leaks.
Checklist to map your unit economics and spot leaks
Use this checklist to build a defensible unit model. I run through it in the first 48 hours on any subscription diagnostic.
Practical levers to plug a 20% margin leak
Once you identify the leak, you need focused levers. Here’s what I usually recommend and why it works.
Example unit economics table (simplified)
| Metric | Value |
|---|---|
| ARPU (monthly) | £84 |
| Payment fees | £2.80 |
| Hosting & infra | £4.00 |
| Support & onboarding (avg) | £15.00 |
| Refunds (per user) | £2.52 |
| Failed payment losses | £5.00 |
| Gross margin / month | £54.68 |
| Average lifetime (months) | 25 |
| LTV | £1,367 |
Note: This table is simplified — include cohort ARPU, churn, and costs if your business has meaningful variation by acquisition source or tier.
How to present this to leadership
I recommend a one-page executive view: headline LTV, headline CAC, the % margin leak with the top three causes, and the top two actions with expected upside (e.g., recover 50% of failed payments = +7% to margin). Use visuals: cohort LTV heatmaps, waterfall charts that start with gross margin and subtract each leak to show the wedge. People understand a waterfall quickly.
When I present this, I always include a pragmatic timeline: quick wins (billing fixes, dunning) in 30 days, medium changes (automation, onboarding redesign) in 90–120 days, and structural changes (pricing or packaging) in the next 6–12 months.
If you want, I can give you a starter spreadsheet template that builds this model with editable assumptions and a waterfall chart. It’s the same one I use in first diagnostics and it will help you find that 20% leak quickly.
