Bottom up
DCFForecasting revenue from the company's own operating units: stores, customers, contracts, capacity.
A bottom up forecast builds from what the business actually consists of. A retailer forecasts store count multiplied by sales per store. A software company forecasts customers multiplied by average revenue per customer. A manufacturer forecasts capacity, utilisation and price.
It is far more defensible than a top down build because each input is something management can be asked about directly, and each can be checked against history. It also makes the forecast auditable: if revenue is wrong, you can see which driver was wrong.
It is the natural basis for scenario analysis, since you can flex one operational driver at a time rather than moving an abstract growth rate.
Its limit is the market context. A bottom up build can quietly assume the company grows far faster than the market it sells into, which is why the two approaches are best used together and reconciled.
Worked example
A retailer has 120 stores averaging 4.2 of revenue each, so 504 today. It opens 15 a year for three years, with new stores maturing to 3.6 in year one.
Year three revenue is the existing estate growing at like for like plus the contribution of 45 newer stores, each of which can be sanity checked against actual store data.
Every input here is something management can be asked about directly, which is why this build survives scrutiny better than a growth rate.