Small Word,  Big Definition
A process that will help prepare you for an upcoming scenario. Using past data (History, Trends) to reduce risks due to uncertainty.   Here are my simple approach to making a forecast (in Business/Marketing/Sales)

Sales forecast, is a prediction of sales value over a period of time. This is the basis of marketing mix and sales planning.
Short Term: Usually a one year period where sales budget is linked to it thus giving an overall picture for the firms performance. This gives a picture to plan sales resources and prepare expenditure required to meet the sales performance. The helps with assessing cash flow, in / out. The needs and sources.

Long Term: Usually a 5 years’ forecasts. The focus is on Capital Budget needs and process of the firm. It provides for the changing the market strategy of the firm. It includes references to emerging product that the market needs. New market segments that needs to be created, reviewing distributing networks, promotional programs. Reorganizing sales force, marketing setups.

Methods of Sales Forecasting:
There are five methods:
1) Executive Judgment
Based on experiences, past performances, intuitions. Works well when the market is stable. Salesforce composite method and jury of executive opinion are the two popular forms used in this method of sales forecasting.

2) Survey:
Prediction can be ascertained by collecting a response from a survey.
Customer Survey: to determine the changes in tastes and preferences, types and value of price opinions of customers.
Sales force Survey: to determine territory take offs, company’s market share, competitor’s analysis from your sales channel/team.
Dealers are also included in sales force survey.
Expert Survey:  to determine the survey reports from the view point of the industry experts and consultants. This helps identify new dimensions for consideration of managements.

3) Time Series Analysis:
Helps locate trends, seasons, cyclical and random factor changes associated with the past data.

Experience reveal that time series analyses for sales forecasting are quite accurate for short and medium term forecasts and more so when demand is stable or follows the past behavior.

4) Correlation and regression methods:
The method examines the past sales volumes with one or more other variables.
The variables considered are usually population, per capita income, or gross national product.  The regression analysis is carried out to compare sales with that of changes in economics, competition, and internal variable of the firm.  The goal is identifying the association of factors that influences performance.

Advance version will have:
Cause and Effect Relationship.
Econometric model,
Input-Output Model,
Life-Cycle Analysis,
New Product’s Growth rate on S-Curve

5) Market Tests:
Used to develop one time forecasts particularly to new products. This gives insight to various elements of marketing mix vs consumer’s actual purchases and responsiveness of the product.
Experiences puts out that combining various methods results greatly surpasses most individual methods of sales forecasts. Combined use of quantitative and qualitative methods of sales forecasting in a given situation rather than using either two can improve accuracy of the forecasts.

Here is a 1971 Review Article that was published in Harvard Business Review. (Click Here)