Revenue is the top of the P&L and therefore the top of the DCF. Every number that follows is some function of it. Get the revenue forecast wrong and the model is wrong, full stop.
Top-down vs bottom-up
A top-down forecast starts with the addressable market and the company's market share. 'Kenyan banking deposits will grow at 12% a year; KCB will hold 17% market share, up from 16%.' Useful for a smell test, dangerous as a primary forecast because the assumed market share is usually too generous.
A bottom-up forecast starts with the unit drivers. Customers × ARPU. Stores × revenue per store. Tonnes × price per tonne. Subscribers × monthly fee × 12. Bottom-up forecasts force you to engage with the operations of the business and are far harder to fudge. The discipline of building one will tell you whether your top-line story is even arithmetically possible.
The 5-year explicit period
Standard practice is a 5-year explicit forecast because that is roughly the horizon over which the business model is recognisably the current business. Beyond 5 years, you are predicting macro and competitive dynamics no one can see. The terminal value picks up everything from year 6 to forever.
S-curves, fade, and convergence
Every fast-growing business eventually slows down. Pricing power gets competed away, market saturates, growth converges to GDP-plus. A revenue forecast that compounds at 25% for 10 years is almost certainly wrong; the question is when it slows, not whether. The 'fade' assumption — how revenue growth decelerates over the explicit period — is one of the highest-leverage choices in the model.
Sanity-check against the market
If your year-5 revenue implies a market share materially higher than today's, ask why competitors will let that happen. If your forecast revenue exceeds the size of the relevant market in real terms, you have made an error. Plot your forecast against the historical 5-year actuals — does the slope look the same, faster, or slower? Defend the difference.
Currency, inflation, and Kenya-specific concerns
Kenyan-listed companies often report in nominal KES. A 15% revenue growth forecast in a 7% inflation environment is only 7-8% real. When comparing across years and across countries, decide upfront whether you are forecasting in real or nominal terms — and use a discount rate consistent with that choice. Mixing real and nominal is the most common, and most easily missed, error.
Exercise
You are forecasting revenue for Safaricom over a 5-year DCF horizon. The business has three revenue lines: voice/SMS (legacy, structurally declining), mobile data (growing strongly), and M-Pesa financial services (growing very strongly). Each has different unit drivers. (1) Build the bottom-up forecast structure for each line — what are the right unit drivers? (2) Why is the top-down 'company revenue × growth rate' approach wrong for Safaricom specifically? (3) Year 1 actuals: voice/SMS KES 50bn (-8% YoY), data KES 70bn (+15% YoY), M-Pesa KES 130bn (+18% YoY). Sketch reasonable year-5 forecasts and explain the fade assumptions for each line.