How to Forecast Revenue: Methods and a Simple Model

How to forecast revenue starts with breaking revenue into the drivers that create it, estimating each from your own history, and multiplying them through, instead of applying a growth percentage to last year's total. A driver-based revenue forecast is easier to explain, easier to challenge and easier to improve than a top-down guess, because every assumption is visible. This guide builds a simple revenue model for a product or service business with repeat customers, shows the arithmetic, phases it by season, tests the sensitivity and lists the checks to run before you share it. It is part of budgeting and forecasting and leads into sales forecasting methods and pipeline-based sales forecasting.

Why drivers beat a growth percentage

A growth percentage ("plan 5 percent more than last year") hides what has to happen: more customers, more orders per customer, higher prices, or all three. When the result misses, you cannot tell why. A driver model replaces one guess with several smaller, testable ones.

Choose the right drivers

The drivers depend on the business model.

Pick drivers that you can measure, that you can influence or at least observe, and that explain most of the variation.

A worked example

A hypothetical business has 12,000 customers who bought last year, and revenue last year of $2,700,000. The management team builds next year's revenue from repeat customers and new customers.

Inputs

The inputs come from the company's own order history: retention measured as last year's buyers who bought again, order counts per customer from the same data, and new-customer orders from last year's new cohort. Where history is unavailable, treat the input as an assumption and flag it.

Calculation

Repeat customers = 12,000 × 80% = 9,600 Repeat orders = 9,600 × 3.0 = 28,800 Repeat revenue = 28,800 × $85 = $2,448,000

New orders = 3,000 × 1.8 = 5,400 New revenue = 5,400 × $70 = $378,000

Total revenue forecast = $2,448,000 + $378,000 = $2,826,000

Growth over last year = $2,826,000 / $2,700,000 - 1 = 4.7%.

The model says where the growth comes from: new customers add $378,000 and repeat customers contribute most of the base. A 4.7 percent increase is not an arbitrary target; it is the result of stated assumptions.

Phasing by season

Revenue is rarely even through the year. If last year's quarterly sales shares were 22, 24, 23 and 31 percent (a holiday-quarter peak), apply them to the annual total:

Check: 621,720 + 678,240 + 649,980 + 876,060 = 2,826,000. Use at least two years of history to confirm that the pattern is stable. See adjusting forecasts for seasonality.

Test the sensitivity

Change one driver at a time to see which matters most.

Calculations:

• Retention at 75%: repeat customers 9,000; orders 27,000; revenue 27,000 × $85 = $2,295,000. Total = $2,295,000 + $378,000 = $2,673,000, which is $153,000 lower.
• New customers at 2,400: new orders 2,400 × 1.8 = 4,320; revenue 4,320 × $70 = $302,400. Total = $2,448,000 + $302,400 = $2,750,400, which is $75,600 lower.
• Repeat order value at $80: repeat revenue 28,800 × $80 = $2,304,000. Total = $2,304,000 + $378,000 = $2,682,000, which is $144,000 lower.

Retention has the largest effect here: five points of retention are worth $153,000, about twice the effect of a 20 percent shortfall in new customers ($75,600). That tells management where effort pays off: keeping customers matters more than acquiring them, at least in this model. For a fuller treatment, see sensitivity analysis in financial models.

Build it in a spreadsheet

Layout:

1. Inputs sheet: every driver in its own labeled cell, with source and date.
2. Calculation sheet: formulas that reference only the inputs, no typed numbers.
3. Output sheet: annual and quarterly or monthly revenue, with a comparison to last year.
4. Checks: a cell that confirms phased revenue sums to the annual total.

Example formulas, assuming named cells:

Repeat revenue = Customers × Retention × OrdersPerCustomer × AOV_Repeat
New revenue    = NewCustomers × NewOrders × AOV_New
Total          = Repeat revenue + New revenue

Keep the model simple enough that the person who owns each driver can understand it.

Split the forecast by segment

Averages hide differences. If retention, order frequency or order value differ by customer type, forecast each type separately and add them up. Suppose the 12,000 customers are really two groups:

Trade accounts are 17 percent of customers and 63 percent of repeat revenue ($1,620,000 / $2,556,000). Losing ten points of trade retention costs 2,000 × 0.10 × 6.0 × $150 = $180,000, more than the whole five-point retention sensitivity in the blended model. The blended model would have missed that exposure. Segmenting is worth the extra rows whenever one group drives a large share of revenue. (These segment figures are a separate illustration, not a restatement of the blended example above.)

Phase the forecast by month

Quarterly phasing is enough for a high-level plan, but cash, staffing and purchasing run monthly. Apply monthly seasonal shares that sum to 100 percent, check the sum and compare each month with last year. Where a launch, price change or trade show falls in a specific month, adjust that month explicitly and note why, so that the shape of the year can be reviewed line by line.

Roll the forecast forward

A revenue forecast made in December is stale by April. After each month closes:

1. Replace the forecast with actuals for the closed month.
2. Re-estimate the drivers that moved (for example, retention tracking below plan).
3. Re-forecast the remaining months.
4. Record the change in the full-year number and the reason for it.

This habit turns the forecast into a rolling forecast and prevents the surprise of discovering in October that the plan is off by 8 percent. See rolling forecasts for the process.

Link the forecast to the plan

Revenue is the start of the budget. Once it is agreed, variable costs follow volume, and fixed costs follow decisions. The forecast also drives cash: revenue is not cash until customers pay. See cash flow management.

Check the forecast before you share it

Improving the forecast over time

• Save each version and compare with actuals monthly.
• Refine the drivers whose errors are largest.
• Measure bias. See forecast bias in financial planning.
• Add detail only where it improves decisions, such as splitting customers by segment if retention differs.

Link to unit economics

Revenue forecasts are most useful when connected to the margin behind them. A revenue plan that depends on discounting to win new customers may add revenue and subtract profit. Connect each customer group to its contribution margin. See unit economics.

Common mistakes

• Applying a growth percentage with no drivers.
• Using drivers you cannot measure.
• Forgetting churn or retention.
• Ignoring seasonality.
• Single-point answer with no range.
• Treating new customers and repeat customers as the same.
• Hard-coding numbers inside formulas.

FAQs

How do you forecast revenue?

Break revenue into drivers such as customers, orders and average order value, estimate each from history, and multiply them through, then phase by season and test sensitivity.

What is the simplest revenue forecasting method?

Last year plus a growth percentage, but a driver-based model is more transparent and more useful.

How do I forecast revenue for a small business?

Use a spreadsheet with an inputs sheet of drivers (customers, orders, price), a calculation sheet and a monthly output, and update it with actuals.

How accurate should a revenue forecast be?

It depends on the business and horizon. Measure error against a naive benchmark and aim to improve. See the accuracy guide for measures.

What data do I need?

At least a year, preferably two or more, of sales by product, customer and period.