CashFlowABC Episode 4 – Forecasting

Podcast: Play in new window | Download

A forecast is an estimate of future demand based on market indicators and/or past performance plus market indicators.   We forecast to reduce the amount of uncertainty in order to prepare for demand.

Who is impacted by a forecast?

• Finance
  • Investment capital
  • AP forecasting
  • Operational budgets
• Inventory planning – for make to or order to stock environments.
• Capacity planning – to ensure forecast does not exceed the ability of the company.
• Procurement – to obtain the raw materials or products to support the forecasted sales. The finished goods forecast is broken down further into the materials planning requirements through the bills of materials.
• Operations
• Investors

Issues you need to consider

• Data
  • Quantitative
    • Seasonality
    • Sales history
      • Record “actual” demand not sales history.
  • Qualitative
    • Expert opinion
    • Customer groups
    • Market estimates
• Supply availability
  • Supplier or production issues.
• Capacity
• Available Capital
• Current market conditions
• Frequency of the forecast
• Your planning horizon
  • Your longest lead time item will determine your planning horizon.
  • Aggregate forecasts are more accurate.
    Determine detail through product families and planning bills.  Below is an example of a product family planning bill:
    
  • Planning bills use individual item and family demand history to breakdown an aggregate forecast into individual item requirements.

Measure your forecast performance.

• Rate of forecast consumption.
  • Divide your forecast into smaller periods to track the consumption.  As an example, divide a weekly forecast into daily checks.  If the forecast was 100 for the week you should sell 20 per day if your demand is flat.  Take into consideration any increases or decreases based on your historical demand or known forecast indicators.
  • If possible adjust the forecast to match the new rate of consumption.
• Compare forecast to actual demand.
  • Track forecast accuracy by period.
  • APE (Absolute Percentage of Error) – The absolute of ((actual demand – forecasted demand) / actual demand) x 100.
    • Does not consider the direction of the variance.  Used to calculate the absolute variance for a single period.
  • MAPE (Mean Absolute Percentage of Error) – The average of the APE across a range of periods.
  • MAPE (Mean Absolute Percentage of Error) – The average of the APE across a range of periods.
    Does not consider the direction of the error.  Used to calculate the absolute variance across a range of periods. Below is an example of the APE and MAPE.
    
  • MAD (Mean Absolute Deviation) – The absolute of (actual demand – forecasted demand) / # of periods.
  • Does not consider the direction of the error.  Used to calculate the absolute variance across a range of periods.
  • Also is approximately the standard deviation / 1.25.
  • Standard Deviation – The square root of the absolute of the actual demand – the forecasted demand squared / the number of periods.
  • Does not consider the direction of the error.  Used to calculate the absolute variance across a range of periods.
    Also is approximately the MAD x 1.25.  Below is an example of the standard deviation equation:
    
  • Bias – Forecast bias is a consistent variation from the average (mean) in one direction.  The goal is a bias of zero.  This demonstrates that any variance is due to normal fluctuation and is not attributable to the method or data.  In the example above the variance total is 700.  That divided by the number of periods is +54.  This is a positive bias and you need to determine why the forecast is higher than the actual demand so you will understand the reasons for lowering your forecast or why demand has fallen short of forecast. 

   

   

   

• Understand the inaccuracy.
• Reasons for forecast inaccuracy
  • Lack of participation
  • Inaccurate data
    • Qualitative
      • No hedging or “sandbagging”.
    • Quantitative
  • Incorrect method
  • Inappropriate data
  • Insufficient data
  • Lack of monitoring
  • Oops
• Apply what you have learned.

 

 

 

Forecasts are nothing more than educated guesses.  And, as human beings, we do best what we do most often.  So that being said the more often you follow proper forecasting processes, understand that it won’t be “perfect”, understand your variances and apply what you have learned to your next forecast your forecast will only get easier and more accurate.  This will help you service your customers and keep costs down.

– David