By Conrad Carlberg

Part of Excel Sales Forecasting For Dummies Cheat Sheet

The Data Analysis add-in, formerly known as the Analysis ToolPak, enters formulas on your behalf so that you can concentrate on what’s going on with your data. It has three different tools that are directly useful in forecasting — Moving Average, Exponential Smoothing, and Regression — as well as several others that can be of help. Here’s a list of some of the tools that are part of the Data Analysis add-in.

Tool What It Does
ANOVA There are actually three different ANOVA tools. None is specifically useful for forecasting, but each of the tools can help you understand the data set that underlies your forecast. The ANOVA tools help you distinguish among samples — for example, do people who live in Tennessee like a particular brand of car better than those who live in Vermont?
Correlation This tool is an important one, regardless of the method you use to create a forecast. If you have more than one variable, it can tell you how strongly the two variables are related (plus or minus 1.0 is strong, 0.0 means no relationship). If you have only one variable, it can tell you how strongly one time period is related to another.
Descriptive Statistics Use the Descriptive Statistics tool to get a handle on things like the average and the standard deviation of your data. Understanding these basic statistics is important so you know what’s going on with your forecasts.
Exponential Smoothing This tool’s name sounds ominous and intimidating, which the tool is not. When you have just one variable — something such as sales revenue or unit sales — you look to a previous actual value to predict the next one (maybe the previous month, or the same month in the previous year). All this tool does is adjust the next forecast by using the error in the prior forecast.
Moving Average A moving average shows the average of results over time. The first one might be the average for January, February, and March; the second would then be the average for February, March, and April; and so on. This method of forecasting tends to focus on the signal (what’s really going on in the baseline) and to minimize the noise (random fluctuations in the baseline).
Regression Regression is closely related to correlation. Use this tool to forecast one variable (such as sales) from another (such as date or advertising). It gives you a couple of numbers to use in an equation, like Sales = 50000 + (10 * Date).