Certain data is required to generate an accurate baseline; with additional data being helpful to add to the quality of the analysis.  The following is a summary of the types of data recommended. 

The data falls into 5 groupings:

  1. Sales data.  Units of weekly sales including price. 
  2. Promotion data.  Dates of the promotions and other relevant information. 
  3. Marketing variables (market conditions).  Major factors that impact sales such as strikes, competitive activity, weather, coupons and end of year sales incentives.
  4. Corporate Calendar.  This is necessary for aggregation into monthly, quarterly or annual numbers and needs to reflect the calendar that is in use by the Company. 
    1. Type of calendar - typical calendars are months containing 5 weeks of sales then two months of 4 weeks each or a 13-month calendar with 4 weeks in each month
    2. Fiscal start and end dates.  Does the year always end on a set day (e.g. December 31) or do days roll over into the next year when a 52 week 7 days/week calendar is used? 
    3. Holidays.  This information is necessary (particularly for shipment data) as the number of shipping days in a month can materially impact the comparison from year to year even when daily sales are the same.
  5. Control files, input files and other constraints file.  These are only for the more sophisticated and experienced users and will be covered in more detail in the Input/Constraints file paper.

Sales data requirement is obvious.  Less clear is the period of data.  Only weekly data makes sense since promotions are run weekly.  Weekly data is also essential for determining such numbers as how much of sales was trade/pantry loading.

 Promotion data is used to determine when a promotion was being offered.    Sometimes that is all a user has and can be used as a minimum level of data.  Some of the more sophisticated models calculate the promotion dates from the sales data.  Very often those models are more accurate in determining when promotions are run because historical data can have errors.  The promotion data is the most complex data file since it can be expanded to include numerous measures of costs (e.g. discount, free goods, coupons, trade payments), type of promotion (e.g. off-invoice, free goods, on pack promotions, IRC’s), and marketing factors to evaluate why some promotions are more successful than others (e.g. level of discount, length of promotion.  This file may also contain the results from evaluating the promotion. 

Marketing Variables.  Many types of initial analysis do not use this data.  It is usually not necessary for the unsophisticated user as it requires an understanding of the factors that impact sales beyond seasonality, trend, and promotions.  It also requires the collection of data and using that in a model.  For example, if weather is a factor (e.g. soda, water, beer) then the historical weather data must be collected and built into a data file.  This is often not something an unsophisticated user will do particularly if the value of the decisions arising from the analysis is not high.