Forecasting Revenue Growth#

The most important input in a DCF model is revenue—not because it directly appears in Free Cash Flow, but because almost everything else flows from it. Margins, working capital, and capital expenditures are often projected as percentages of revenue. Get revenue wrong, and the entire valuation falls apart.

Starting Point: Historical Analysis#

Before projecting forward, look backward. A company's past growth reveals patterns, constraints, and possibilities.

Calculating Historical Growth Rates#

Let's analyze a fictional company, TechCorp:

Compound Annual Growth Rate (CAGR):

CAGR = (Ending Value / Beginning Value)^(1/n) - 1
CAGR = ($1,467M / $800M)^(1/5) - 1 = 12.9%

But the simple average of yearly growth is 13.0%. Which should we use?

Reading the Story Behind the Numbers#

TechCorp's growth tells a story:

• 2020-2021: Strong growth (20%) during expansion phase
• 2022-2023: Growth deceleration (15% → 10%) as the company matures
• 2024: Significant slowdown (5%) suggesting market saturation

This pattern—rapid early growth that fades—is nearly universal. Companies don't grow at 20% forever.

Two Approaches to Forecasting#

1. Top-Down: Start with the Market#

Top-down forecasting begins with the total market and works down to the company.

The TAM/SAM/SOM Framework:

Example: CRM Software Company

TAM: $50B (all CRM software globally)
SAM: $15B (mid-market, North America + Europe)
SOM: $1.5B (realistic capture in 5 years)

Current Revenue: $500M
SOM Implies: 3× growth over 5 years
Annual Growth Rate Needed: ~25% per year

Is 25% annual growth reasonable? That depends on:

• Market growth rate (is CRM itself growing?)
• Competitive dynamics (is the company gaining or losing share?)
• Product strength (reviews, customer retention, new features)

2. Bottom-Up: Build from Company Data#

Bottom-up forecasting aggregates detailed company metrics into revenue.

For a SaaS Company:

Revenue = Customers × ARPU (Average Revenue Per User)

Year 1:
- Starting customers: 10,000
- New customers: 3,000
- Churned customers: (1,500)
- Ending customers: 11,500
- ARPU: $5,000
- Revenue: $57.5M

For a Retailer:

Revenue = Stores × Revenue per Store

Year 1:
- Existing stores: 500
- Same-store sales growth: 3%
- New stores: 25
- Average revenue per new store: $2M (ramping)
- Average revenue per existing store: $3.1M
- Revenue: $1.6B

Building a Revenue Forecast#

Let's build a 5-year revenue forecast for TechCorp:

Step 1: Establish Base Case Assumptions#

Step 2: Project Each Year#

Rather than assuming constant growth, model the trajectory:

5-Year CAGR: 6.0% (realistic given the slowdown)

Step 3: Sense-Check Against Market#

At $1,962M in 2029, where is TechCorp in its market?

• If TAM grows from $15B to $20B (6% CAGR), TechCorp goes from 9.8% to 9.8% market share
• Flat market share is reasonable for a mature company
• This passes the sanity test

Growth Fading: A Critical Concept#

No company grows faster than its industry forever. Eventually, all growth rates converge toward GDP growth (2-3%) or industry growth.

The Growth Decay Pattern#

Modeling Growth Fade#

Here's a systematic approach:

1. Years 1-2: Use management guidance or analyst estimates
2. Years 3-5: Fade toward industry growth rate
3. Years 6-10: Fade toward terminal rate
4. Terminal: 2-4% (nominal GDP growth)

Example Fade Schedule:

Common Forecasting Mistakes#

1. Hockey Stick Projections#

Management teams often show projections that dip then soar—the famous "hockey stick." Be skeptical. If growth has been declining, extrapolate the decline.

2. Ignoring Competitive Response#

If your forecast assumes massive market share gains, ask: won't competitors respond? Price cuts, new features, and acquisitions can derail even the best growth plans.

3. Confusing Revenue Types#

Different revenue types have different characteristics:

4. Forecasting Too Far Out#

The further out you forecast, the less reliable it becomes. Beyond 5-7 years, you're essentially guessing. This is why terminal value matters so much—it captures everything past your explicit forecast.