Revenue forecasting is the process of estimating how much revenue your business will generate over a future period, using the data you have today. There are five methods founders rely on: run-rate, growth-rate extrapolation, top-down, bottom-up (driver-based), and weighted pipeline. Each answers "how much revenue next period?" from a different starting point, and each has a formula. This guide gives you the formula and a worked example for all five, then shows which one to use at your stage.
A revenue forecast is not a promise. It is a defensible estimate built from explicit drivers, so that when the number turns out wrong (it always is, by some amount) you know exactly which assumption to fix. This page is about the forecasting methods themselves. For the full projection package (revenue plus expenses, profit, and cash over three to five years) see SaaS financial projections, and for the underlying metric see revenue growth rate.
The five revenue forecasting methods
| Method | Starts from | Best when |
|---|---|---|
| Run-rate | Your latest period, annualized | You need a fast baseline and revenue is stable |
| Growth extrapolation | A recent growth rate, projected forward | You have a few periods of consistent history |
| Top-down | Market size and a target share | Pre-revenue or a sanity check |
| Bottom-up (driver-based) | Your funnel and subscription drivers | The forecast you actually present and run on |
| Weighted pipeline | Open deals and their win probability | Sales-led motion with a maintained CRM |
1. Run-rate forecasting
The fastest method: take your most recent period and annualize it. For a subscription business you annualize MRR into ARR.
Annual run-rate = latest period revenue × periods per yearMonthly MRR of $50,000 × 12 = $600,000 ARR
Run-rate is a snapshot, not really a forecast: it assumes next year looks exactly like your last month. It ignores growth, churn, and seasonality, so it overstates for a shrinking business and understates for a fast-growing one. Use it as a starting line, then apply one of the methods below.
2. Growth-rate extrapolation
Take a recent growth rate and project it forward. The straight-line version applies one rate repeatedly (compounding), which is the same math as CAGR.
Forecast = current revenue × (1 + growth rate)periods$50,000 MRR growing 8% per month for 12 months: 50,000 × 1.0812 = $125,908
The trap is obvious once you see the number: 8% a month compounds to roughly 152% growth over a year. Real growth rates decay as you get bigger, so extrapolating a single early rate produces the classic hockey stick. If you use this method, taper the rate down over time rather than holding it flat.
3. Top-down forecasting
Start from the size of the market and assume you capture a share of it. Size the market properly first with TAM, SAM, and SOM, then apply a target share to your serviceable segment.
Forecast = serviceable market (SAM) × target market share %$2,000,000,000 SAM × 0.5% target share = $10,000,000
Top-down is easy to inflate ("we only need 1% of a huge market"), which is exactly why investors distrust it as a primary forecast. Its real job is a sanity check: if your bottom-up build implies a share of the market that is implausible, one of your assumptions is wrong.
4. Bottom-up (driver-based) forecasting
This is the method to build your real forecast on, and the one investors expect. Instead of assuming a share of a market, you build revenue up from the levers you actually control: how many customers you add, what they pay, and how many you keep. For SaaS, that is the MRR build.
Ending MRR = starting MRR + new MRR + expansion MRR − churned MRR$50,000 + (200 new × $100) + $3,000 expansion − $4,000 churn = $69,000
New MRR itself comes from your funnel: visitors, trial or lead conversion, and average price, which ties the forecast back to acquisition spend you can plan. Roll the build forward month by month and you have a revenue forecast where every number traces to a driver an investor can challenge. Feed it through the growth funnel and you can forecast the marketing budget required to hit the number, not just the number.
Bottom-up takes more work than the other methods, but it is the only one that tells you what to do differently when the forecast and reality diverge. Change churn or conversion, and the whole forecast moves the way the business would.
5. Weighted pipeline forecasting
For a sales-led motion, forecast from open deals weighted by their probability of closing. Assign each stage a win rate and sum the expected values.
| Deal | Value | Stage win probability | Weighted value |
|---|---|---|---|
| A | $40,000 | 80% | $32,000 |
| B | $60,000 | 50% | $30,000 |
| C | $100,000 | 20% | $20,000 |
| Forecast (sum of weighted values) | $82,000 | ||
Pipeline forecast = Σ (deal value × stage win probability)$32,000 + $30,000 + $20,000 = $82,000
Weighted pipeline is only as good as your CRM hygiene and the honesty of your stage probabilities. It works for near-term (this quarter) sales-led revenue; it does not replace a driver-based model for the full year.
A note on statistical methods
Larger companies with clean history also use quantitative methods: moving averages, time-series analysis, and linear regression that ties revenue to one or more drivers. These add rigor once you have enough historical data points, but they need that history to be reliable. Most early-stage companies do not have it yet, which is why the five methods above dominate at the founder stage.
Which method should you use?
| Situation | Use |
|---|---|
| Pre-revenue, pitching a market | Top-down for the sanity check, bottom-up for the plan |
| A few months of data, need a quick view | Run-rate, then growth extrapolation with a tapering rate |
| Product-led SaaS, building the real model | Bottom-up (driver-based) MRR build |
| Sales-led, forecasting the quarter | Weighted pipeline for near-term, bottom-up for the year |
| Later-stage with clean history | Bottom-up plus a statistical cross-check |
The strongest forecasts run two methods and reconcile the gap. If your bottom-up build and your top-down sanity check land far apart, that gap is the conversation worth having before you present anything.
Common revenue forecasting mistakes
- Holding one growth rate flat. Compounding an early rate forever is the hockey stick. Taper it.
- Forecasting revenue but forgetting churn. Gross new revenue is not net revenue. Always net out churn.
- One number, no scenarios. Present base, best, and worst. The downside case builds more trust than the upside.
- Top-down as the headline. "1% of a $10B market" is not a forecast. Lead with bottom-up.
- Never updating. A forecast is a living document. Reforecast as actuals arrive and note why you missed.
How far out and how accurate?
Forecast the near term in detail and the long term in broad strokes. Show the first 12 to 24 months monthly (this is where cash and hiring decisions live) and later years annually. On accuracy, near-term forecasts should land within a tight band; anything beyond 18 to 24 months is directional, not precise. The goal is not to be exactly right, it is to be roughly right for a reason you can defend and correct.
Building this by hand in spreadsheets is where it breaks: formulas drift, scenarios multiply into unmanageable tabs, and the model stops matching reality. Adlega builds a driver-based revenue forecast (with MRR, expansion, churn, and cash) from your assumptions, and its AI CFO explains every number back to the driver that produced it. For the complete picture, see the SaaS financial model guide.
Revenue forecasting FAQ
What is revenue forecasting?
Revenue forecasting is estimating your future revenue over a defined period using current data and explicit assumptions. It is built from a method (run-rate, growth extrapolation, top-down, bottom-up, or weighted pipeline) rather than a guess, so each number traces to a driver you can defend.
How do you forecast revenue for a startup?
Build it bottom-up. Start from your funnel (visitors, conversion, price) to project new customers and MRR, net out churn, add expansion, and roll it forward month by month. Use a top-down market estimate only as a sanity check, and present base, best, and worst-case scenarios.
What is the difference between a revenue forecast and a revenue projection?
In practice they are used interchangeably. Where people distinguish them, a forecast is the near-term, method-driven estimate you manage against, while a projection is the longer-horizon output presented in a business plan. Both are the result of running a financial model.
What is the most accurate revenue forecasting method?
For a company with real drivers, bottom-up (driver-based) forecasting is the most accurate and the most defensible, because it ties revenue to levers you control. Weighted pipeline is strong for near-term sales-led revenue. Top-down is the least accurate and belongs as a cross-check, not the headline.
How far into the future should you forecast revenue?
Three to five years is standard for a plan, but only the first 12 to 24 months should be monthly and detailed. Beyond about two years the forecast is directional. Forecasting month 40 to the dollar signals a model that does not understand its own uncertainty.
What is run-rate revenue?
Run-rate annualizes your latest period, for example monthly MRR times 12 to get ARR. It is a fast baseline that assumes no growth or churn, so treat it as a starting point rather than a forecast.
Related: SaaS financial projections, revenue growth rate, cash flow forecasting for SaaS, and the SaaS financial model guide.