# Sales Forecasting Tools by epn46924

VIEWS: 46 PAGES: 3

• pg 1
```									Sales Forecasting Models

I.    All businesses make some kind of sales forecast.

a. Implicit forecast

i. Owners develop of notion of expected sales from past experience

ii. Most likely in small businesses

b. Explicit forecast

i. Forecasts are made and distributed to managers

ii. Follows a defined process

iii. An survey on sales forecasting from a fairly diverse population of
companies reveled

   49% companies used simple tools like Excel, Access for demand
forecasting

   9% did not use any software

   42% used mathematical models

   93.5% of companies that used some forecasting model achieved
significant benefits

   The percentage of companies having forecasting error less than
10% increased from 25 to 67 on use of forecasting models.

II.   Types of explicit forecasts

a. Executive Forecasts

i. Team of company experts meets to formulate forecast

ii. May utilize outside consultant to direct process

iii. Desirable to draw from all segments of the business

b. Delphi Method

i. Utilizes experts as in Executive Forecast
ii. May use external experts

iii. Each member makes individual forecasts

iv. Then is shown forecasts of others anonymously

v.   Revise forecasts in light of feedback from others

vi. Keep this up until process converges on some result

c. Time Series Analysis

i. Use time series model to historical trend data

ii. Example: http://www.futuretoolkit.com/

iii. Look sophisticated and involve sophisticated math techniques

iv. Work very well with data that continues to follow past trends
regardless of how complex

v. Shortcoming is not real information about what drives the sales
trends or cycles

d. Regression techniques

i. A dependent variable, usually sales, is forecast using a set of
explanatory variables each of which contributes to sales

ii. Choice of explanatory variables is critical

1. Variables should meet the logic test and not rely on
correlation to determine which variables are included

2. Data mining

a. Process of testing dependent variable against
hundreds to thousands of possible explanatory
variables

b. Build model using those are have the strongest
correlation

c. Need to be careful using this technique
d. Example

i. Suppose you want to forecast weekly banana
sales by county.

ii. You discover, by data mining, that in countries
where people own a lot of French Poodles they
also consume a lot of bananas

iii. Should you include French Poodles per capital
in model as an explanatory variable for banana
sales?

iii. Linear regression often used when using aggregate data such as
zipcode demographics

iv. Logistic regression often used when using individual data

1. Gives probability a consumer with a set of characteristics will
characteristics data

2. If people with personal characteristics A have a 5%
probability of buying the product then, if you approach 1000
type A people you will make 50 sales.

3. Example targeted direct mailings

4. http://www.extenza-