Sales forecasts are by nature imperfect. But experts say there are ways to squeeze more value out of the projections you're making.
Any good business will have a system of sales forecasting as part of its critical management strategy. But most sales forecasts are, by nature, inexact. The trick, experts say, is to know in which direction they're wrong, and turn that into a picture of how your business is doing.
"People think the forecast is good or bad depending on how accurate it is," says Tim Berry, president of Palo Alto Software, which creates business-planning software and is—despite its name—based in Eugene, Oregon. "What I think is it's how well it breaks into meaningful assumptions you can look at later."
For a small business—say, a restaurant—those assumptions could mean mapping activity in the dining room and keeping track of how many meals are sold at certain times of day. For a larger business, that could mean plotting all activity across departments to see how your products are matching up to industry standards. There's a few ways to test the your sales forecasting to know whether you're getting an accurate read or just dabbling in expensive soothsaying.
1. Use separate numbers. One of the biggest misconceptions about forecasting is that there's one set of numbers that represents the "truth" for your business. In reality, multiple forecasts are necessary in order to represent the needs of different constituencies, says David Stephens, director of sales for Right90, a sales forecasting company based in Austin, which has done forecasting for Sharp and Bivo Networks.
Your sales team might have a forecast designed to meet its number, but product management is interested in the forecast of a specific product, operations is interested in finding out what it needs to produce and when, and finance needs revenue figures, he says. Someone at the top of the ladder needs to be prepared to put all those together and form a cohesive picture.
"Senior management requires the forecast be vetted from all perspectives in order to develop the confidence to make critical decisions," he says.
2. Develop a flexible process. It's impossible to use a single test that will ensure you can track the exact terms, time, and context of every sale. Instead, you should focus on developing a process that can be managed, reevaluated, and modified as conditions change, Stephens says.
"This requires discipline, beginning with ensuring that sales forecasts are updated on a regular basis," he says. That means managers have to understand the sales system, customer history, product delivery, and even the history of the individual salesperson to assess with some certainty the forecast's accuracy.
Big companies often make the mistake of thinking forecasting is just looking at the sales history and taking an average over time. Instead, they need to look at many additional factors as well, says Glen Margolis, president and CEO of Steelwedge Software, which is based in Pleasanton, California, and has forecasted for GE, DirectTV and Sara Lee.
3. Set aside time. Your forecasts won't do you much good if you aren't constantly keeping tabs on them. Berr says it's crucial for companies to set aside a specific time every month (or however often you like) to review the forecasts.
Berry says his company does this every third Thursday of the month: managers bring in lunches, review the data together, and make any work on any high-level decisions that may be called for. It's all part of the broader decision making of the company.
"If there's a set time, everybody involved knows," he says. "You look at, compare and plan for actual results…you start to see management happening."
4. Use a consistent model. Margolis says he believes there's no one model of forecasting that works best for every company. But one efficient method is sometimes one used by restaurant owners: matching this year's sales to last year's and making a guess for the future.
"That to me is the best model," he says. "That empowers people who are actually running business."
But the key is, whatever model is used—whether it is a weighted average over a few months or bare numbers-tracking—needs to be consistently applied over time.
"One of the biggest barriers is people saying 'I'm not qualified to forecast, I didn't take statistics,'" he says. "Well, I do have the degree, and I did take statistics, and still the educated guesses are what really drive the forecast."
Consistent application of the same model standardizes the format, and makes it easier to review year after year.
5. Don't get too complicated. Your business forecasting doesn't have to be a hyper-complicated process that involves high-level mathematics and projections.
"Most businesses are not necessarily very sophisticated," Margolis says. "They don't have a team of statisticians. It's someone with other day-to-day activities who also keeps an eye on forecasting."
Stephens says simple, specific software and applications are available for sales forecasting that provide an audit trail, a history of the forecast, and the ability to align the data with customer relations management. The programs also allow you to note changes to any perspective such as product, territory, customer, or salesperson, he says—much more so than just keeping a spreadsheet on your laptop.
6. Be democratic. If you aren't including all elements of the business in the forecasting process, you are likely to end up with skewed numbers. Margolis says that if people aren't involved in the process, they won't believe the numbers and won't use them, or will fudge the data to fit their personal expectations. A purchasing department, for instance, may up certain numbers so it doesn't run out of stock.
Margolis advocates making forecasting a collaborative process, as much as possible.
"There's ways to manage the collaboration so you're getting the benefit not the downside of it," he says. "Because everyone's participating and they feel like their voice has been heard, they trust it. They're more likely to trust it than to game it."
7. Focus on exceptions. You can tweak the details of the forecast to death, but your main focus should be looking at the exceptions: where the forecast line diverges from actual sales data.
"Don't loose track of the forest for the trees," Margolis says. "You're just constantly trying to improve." Improving forecasting doesn't happen overnight: analysts expect forecasts to include monthly data for about the next 12 months, Berry says, as well as annual data for year two and three. Anything more specific than that is "basically an academic exercise," he says.
Experts say it sometimes takes months of tweaking, adjusting and learning before you can have an accurate guess at how the forecasting will look in the future.
"There's constant elements and dimensions you can refine all the time," Margolis says. "It has to be a core competency, especially in the world we're in today that's very unpredictable."