Forecasting Methods: Which One Fits Your Business?

Start by choosing between past-based and driver-based

Past-based methods extend history. Driver-based methods multiply the factors that create sales. Use history when sales are steady and drivers when conditions are changing.

If your sales have followed a stable pattern for two years, history is a reasonable guide. If you are launching a new product, changing prices or entering a new area, history is a weak guide, and drivers will serve you better.

Whichever you pick, write down the assumptions. When actual results differ, you will know which assumption was wrong. For a full planning walk-through, see financial forecasting for small business.

Run-rate and moving average

A run-rate projects recent sales forward. A moving average smooths out spikes by averaging the last several periods.

Illustrative example: your last three months of sales were $42,000, $48,000 and $45,000. The three-month average is $45,000. A run-rate forecast for the next month would be about $45,000.

• Pros. Fast, easy and reasonable for steady businesses.
• Cons. It lags when sales are growing or falling, and it ignores seasonality.

Use it for a quick sanity check, not as your only method for a longer horizon.

Seasonal and year-over-year methods

Seasonal methods apply last year's monthly pattern to this year's expected total, which suits businesses with predictable busy and slow periods.

If December is 18 percent of your annual sales, and you expect $500,000 this year, December forecast is about $90,000. Adjust for growth or price changes you know about, and note unusual events from last year, such as a closure.

This method works well for retail, restaurants, landscaping and tax services. It needs at least one full year of reliable data. See four ways to fund a seasonal business for managing the slow months.

Driver-based forecasting

Driver-based forecasts build revenue from inputs such as leads, conversion rate, average sale and repeat purchases.

Example for an illustrative service company: 200 leads a month, 25 percent become customers, average job $1,200. That is 50 jobs and $60,000 in revenue. Now you can test changes. What if leads grow 10 percent? What if conversion drops to 20 percent?

The benefit is that the model tells you what to fix. The risk is false precision, so use ranges and update the inputs when real data arrives. Our revenue forecasting article shows variations.

Pipeline and backlog methods

If you sell projects or contracts, forecast from signed work plus a weighted share of proposals you expect to win.

Contractors, agencies and consultants can forecast from the sales pipeline. Weight each opportunity by its likelihood. A $50,000 proposal with a 30 percent chance counts as $15,000. Add signed backlog by the month you expect to bill it.

The pipeline method has a weakness: people are optimistic. Track your actual win rate and use it instead of a hopeful guess. Timing also matters; cash arrives after billing and after the customer pays. See the cash conversion cycle for the delay.

Check accuracy and use the forecast to decide on funding

Compare forecast to actual every month. If you are off by more than 10 to 15 percent, adjust the method or assumptions.

Track the percentage error each month. If you always overestimate, add a haircut. If accuracy is poor in certain months, those months have a factor you have not modeled.

Once you trust the forecast, use it to size funding. Suppose the forecast shows a $40,000 shortfall in month five, ahead of a season that will add profit. A business line of credit or working capital may bridge it. ROI Funder considers 500+ FICO, 6+ months in business and $20,000+ monthly revenue, and applying is a soft credit pull with no obligation.