Financial Modeling Basics: Forecasting Sales

Forecasting Sales is probably the most important step in building a Financial Model and valuing a company. If you get sales very wrong then all your other numbers will be wrong as well, as sales is the key driver of any non-financial services business.

For mature companies there are generally two methods that analysts use to forecast sales:

  • Growth Rate; and
  • Build-up.

Using a growth rate to forecast sales is generally appropriate under the following conditions:

  1. The level of detail you have available about products sold is limited;
  2. The company has demonstrated relatively stable or predictable sales levels that appear linked to an identifiable driver;
  3. The fundamental drivers affecting sales levels are easy enough to comprehend.

To model sales using Growth Rates, you will have one single assumption per period being modelled, which is the growth rate in sales for that year compared to the prior year. Then the formula structure to use is: This Year's Sales = Last Year's Sales x (1 + growth rate)

The Build-up method requires more detail but can deliver greater accuracy and also identifies which particular drivers are causing sales to increase/decrease/remain unchanced.

The Build-up method requires more information than the Growth Rate method. In order to effectively use the build-up method, you need the following information, or at least assumptions around the following information:

  1. You need to identify each product that the company sells;
  2. You need assumptions about the average price each product is sold at throughout the period
  3. You need assumptions about the number of units of each product that are sold throughout the period.

So if a company has 5 products, you need 10 assumptions to forecast sales per period being forecast.

Often companies will disclose some level of detail on the number of products they sell, and the average price and volume of units for each product, however sometimes you will need to make an assumption yourself. For instance if we were forecasting the level of sales for Coca-Cola Amatil in Australia in a year we might use the following build-up (note these assumptions are not verified for accuracy and may misrepresent CCL's actual revenue drivers!):

  • Average person drinks 3L of CokeProducts per week x 20,000,000 relevant customers x 52 weeks in a year = 3,120,000,000L per year;
  • Average revenue per L of Coke product = $1.50
  • Revenue = 3.12B x $1.50 = $4.68B this year

Then we can sanity check this against their historical levels to see whether our assumptions made sense: In 2009 Coca-Cola Amatil had total trading revenue of $4.40B so our forecast assumes an implicit growth rate of 6.4% in trading sales which seems within the bounds of reason.