Financial Forecasting for Small Businesses: A Practical Guide
Forecasting is not about predicting the future correctly. It is about having a written expectation you can compare reality against, so you notice early when things drift. This guide covers the methods small businesses actually use, how to forecast sales, costs and cash, and how to improve your accuracy month by month.
Key takeaways
- Forecast sales from drivers and history, costs from commitments, and cash from timing. They are three different exercises.
- Use a rolling forecast that always looks 12 months ahead, rather than a fixed annual budget that goes stale by March.
- Compare forecast with actuals every month and record the variance. The gaps teach you more than the forecast does.
- Pair a base case with a downside case. If the downside breaks the business, fix the plan before it happens.
A budget is a plan for what you want to spend. A forecast is an estimate of what you expect to happen. Many small businesses have the first and not the second, which is why they discover a slow quarter only when the bank balance drops. A forecast, updated monthly, gives you two or three months of warning instead.
The methods here do not require statistics software. A spreadsheet, twelve months of accounting history and an hour a month are enough for most businesses with under a few million in revenue.
What to forecast, and in what order
| Forecast | Driven by | Main use |
|---|---|---|
| Sales | Customers, orders, pricing, seasonality, pipeline | Everything else depends on it |
| Cost of sales | Sales volume and unit costs | Gross margin and purchasing plans |
| Operating expenses | Commitments, headcount, planned spending | Profitability and hiring decisions |
| Cash | All of the above plus payment timing, debt, tax and capital spending | Avoiding a shortfall |
Always start with sales, because it is the most uncertain and the one everything else scales from. Cash comes last because it is derived from the other three plus timing.
Four ways to forecast sales
1. Run-rate and growth
Take the average of the last three months and apply a growth rate based on the last year. If monthly sales averaged $50,000 and the business grew 15% over the past year, forecast roughly 1.2% monthly growth from a $50,000 base. This is the simplest method and works for stable businesses without strong seasonality.
2. Seasonal pattern
Calculate each month's share of annual sales from the past one or two years, then apply those shares to your expected annual total. A garden center that does 18% of its year in May and 4% in January should not forecast a flat line. If you have two years of history, average the two patterns to smooth out one-off events.
3. Driver-based
Forecast the inputs, then multiply. For an online store: visitors × conversion rate × average order value. For a consultancy: billable staff × billable hours × rate. For a subscription business: opening customers plus new minus churned, times price. Driver-based forecasts are more work but they tell you why a month will be good or bad, which makes the forecast actionable.
4. Pipeline-weighted
For businesses with a sales pipeline, list every open opportunity with its value, expected close month and a probability based on stage. Multiply value by probability and sum by month. If your historical close rate from proposal stage is 40%, a $20,000 proposal contributes $8,000 to that month's forecast. Keep probabilities honest by checking them against what actually closed.
Worked example
Suppose a small marketing agency has three retainers worth $12,000 a month combined, four proposals out worth $30,000 in total at a 40% historical close rate, and a track record of roughly $5,000 a month in small ad-hoc work. The forecast for next month is $12,000 of retained revenue, $12,000 of weighted pipeline (40% of $30,000) and $5,000 of ad-hoc work, which is $29,000. The owner writes that down. When the month closes at $24,000 because two proposals slipped, the $5,000 variance is visible, explainable and useful.
Common mistake
Forecasting sales as "last year plus 20%" because that is the target. Targets are goals; forecasts are expectations. Mixing them produces a forecast that is always missed, which teaches the team to ignore it. Keep the stretch target on a separate line and forecast what you actually expect.
Forecasting costs
Costs are easier because most of them are already committed. Split them into three lines.
- Cost of sales: apply your gross margin percentage to forecast sales, adjusted for any known supplier price changes. If margin was 58% last quarter and your main supplier raised prices 5%, model the effect explicitly.
- Committed expenses: rent, salaries, loan payments, insurance, software subscriptions. Pull these from contracts and payroll. They are the most accurate part of any forecast.
- Discretionary expenses: marketing, travel, training, new hires. Forecast what you plan to spend and when, and treat these as the levers you pull if sales fall short.
Forecasting cash
The profit forecast tells you whether the business is working. The cash forecast tells you whether it will survive the next quarter. The difference is timing. To convert a profit forecast into a cash forecast:
- Shift sales by collection time. If customers pay in 30 days on average, cash from this month's sales arrives next month. Use your actual days sales outstanding, not your stated terms.
- Shift purchases by payment time. Apply your usual payable terms to cost of sales.
- Add items that never touch the profit and loss. Loan principal repayments, equipment purchases, owner drawings, VAT or sales tax collected and remitted.
- Place tax payments in the months they fall due. Estimated tax in the US, corporation tax in the UK and similar payments elsewhere arrive on fixed dates, not spread evenly.
- Calculate closing cash each month. Opening balance plus receipts minus payments. Highlight any month that falls below a minimum you set, for example one month of fixed costs.
For businesses with tight cash, run a 13-week weekly cash forecast alongside the monthly one. Thirteen weeks is a quarter, which is long enough to act and short enough to be accurate.
Rolling forecasts beat annual budgets
An annual budget is set once and compared against for twelve months, even after the first quarter has proved it wrong. A rolling forecast adds a new month every time one closes, so you always see twelve months ahead and every assumption is at most a month old.
The monthly routine takes about an hour:
- Enter actuals for the month that just closed, from your accounting software.
- Record the variance between forecast and actual for sales, gross margin, expenses and closing cash, in dollars and percent.
- Write one line explaining each variance over 10%. "Two proposals slipped to next month." "Supplier price rise." This is the most valuable part.
- Update the assumptions that the variances disproved: close rate, collection days, margin.
- Add the new twelfth month at the end of the forecast.
- Check the downside case still works with the updated numbers.
Worked example
Imagine a small bakery that forecast $38,000 in June sales and closed at $34,500, a 9% miss. The variance note reads: "Two wholesale accounts cut standing orders by 20% from mid-month." The owner updates the wholesale line for the rest of the year, which cuts forecast cash at the December low point from $21,000 to $9,000. That is still positive, but below the $12,000 minimum the owner set. Three months before the problem, the bakery has time to add a wholesale account or trim the holiday hiring plan.
Measuring and improving accuracy
Track forecast accuracy itself. For each month, calculate the absolute percentage error for sales: the difference between forecast and actual, divided by actual, ignoring the sign. Average it over the last six months. An average error under 10% for one month ahead is good for most small businesses; 20% or more means the method or the inputs need work.
Also check for bias. If you miss high six months in a row, your forecasts are optimistic, which usually means the growth rate or close rate is too generous. If you consistently beat the forecast, you may be under-investing because the plan is too cautious.
Pro tip
Forecast in ranges for the months furthest out. A forecast of "$40,000 to $48,000" for month ten is more honest than "$44,350" and makes it easier to set a downside case. Narrow the range as the month approaches.
Good forecasting habits
- Driver-based sales forecast with seasonality
- Monthly variance notes in plain language
- Cash forecast with real collection and payment timing
- A downside case reviewed every month
Habits that break forecasts
- Treating the stretch target as the forecast
- Spreading tax and annual bills evenly across months
- Updating the forecast only when things go wrong
- Ignoring accuracy and bias over time
Your forecasting action plan
- Export the last 24 months of sales and calculate each month's seasonal share
- Choose a sales method (run-rate, seasonal, driver or pipeline) and build a 12-month forecast
- List committed costs from contracts and payroll, then add discretionary plans
- Convert to a cash forecast using actual collection and payment days
- Set a minimum cash balance and flag any month below it
- Book a monthly hour to add actuals, write variance notes and roll the forecast forward
- Track average forecast error and adjust the method after six months
Frequently asked questions
How far ahead should a small business forecast?
Twelve months in monthly detail, rolled forward each month, covers most needs. Add a 13-week weekly cash forecast if cash is tight, and a two- or three-year outline if you are planning a major investment or raising money.
What if I have no history to forecast from?
Use drivers and assumptions, state them clearly and build a wide range. Replace assumptions with actuals as soon as you have them. The first six months of a new business are mostly about learning what the real numbers are.
Should the forecast match the budget?
No. The budget is what you authorized people to spend. The forecast is what you now expect to happen. Comparing the two is useful; forcing them to be equal hides information.
Do I need forecasting software?
A spreadsheet is enough for most small businesses. Software becomes worthwhile when several people need to update the forecast, when you have many products or locations, or when you want automatic links to your accounting system.
Related reading
The bottom line
Forecast sales from drivers and history, costs from commitments and cash from timing. Roll the forecast forward every month, write down why you missed and measure your own accuracy. Within a few months, the forecast will stop being a guess and start being the earliest warning system your business has.
