First computing doctor forecasts the cost of doing

20 September 2007

First computing doctor forecasts the cost of doing business

The first graduate from an innovative computing doctoral programme is helping businesses to use their forecasting software more effectively.

Dr Peter Catt is a Supply Chain Solution Manager at IT consultancy Soltius New Zealand Limited and received his Doctor of Computing degree - the first ever to be awarded the qualification - at Unitec’s graduation ceremony in Auckland this week.

Soltius is a leading implementer of SAP enterprise resource planning (ERP) systems and Dr Catt’s doctoral work looked at the use of these solutions in sales forecasting. He says that forecasting models are an important business tool, but the way in which they are often used can have a negative impact on business profitability.

“SAP ERP solutions are used by many of the world’s most successful companies and have comprehensive forecasting functionality, but most companies don’t use the right models because they don’t know what the real forecasting error costs are.”

This is because most businesses use basic statistics to show the difference between the sales forecasts and actual sales – information that is of little use in aiding decision-making, says Dr Catt.

“Learning that they sold a few percent more or less of a particular product line than the forecast tells a business absolutely nothing. This statistical data doesn’t take into account the margins on a product line or the costs of having safety stock.”

And with many businesses ordering stock from overseas months in advance, getting their predictions right is crucial.

For example, if a business sells 20 percent more of a product than forecasted they would take that into account when making their next order. But if the margin on that product is very low, the cost of having safety stock on hand may be more than the profits realised from the sales.

“Conversely, on a high margin product line the cost of having additional safety stock on hand may be worth it. Traditional forecast error metrics simply do not reveal the financial impact in the real world. And as a result, the business can never be sure if they are using the right forecast model.”

Dr Catt has developed a new approach - the Cost of Forecast Error metric - which he says is a better guide for businesses, not only in calculating forecast error, but in choosing the right IT solution in the first place. “SAP alone has nine different forecasting models and most businesses are using the wrong ones - if they’re not, it’s probably a fluke.

“Running a few simulations using the cost of error metric will quickly tell a business which IT solution to purchase to meet their specific needs.”

His work has already been published in a leading US journal and has had positive responses from a number of US academics, including an MIT professor.

Dr Catt, who has previously completed an MBA from Henley Management College in the UK, says that getting the forecast model right will pay off in the long run. “Businesses have been operating in the dark to a large extent when setting their stock levels and placing orders with their suppliers.

“This cost to forecast error calculation is a relatively simple method that can have a significantly beneficial impact on the bottom line.”

ENDS

Fact sheet:

* Soltius New Zealand - IT consultancy, New Zealand’s largest dedicated implementer of Enterprise Resource Planning (ERP) systems. Almost 70 consultants with offices in Auckland and Wellington

* SAP – giant Germany-based software company, the global leader in the ERP market

* Unitec – Auckland-based tertiary institution, offers New Zealand’s only Doctor of Computing (DComp) programme

* Dr Peter Catt – Soltius Supply Chain Solution Manager, first to receive the DComp, at Unitec’s spring graduation at the Auckland Town Hall this week

* Many ERP systems have sales forecasting model functionalities – the SAP All-in-One solution alone has nine different forecasting models to choose from

* Different product lines require different forecast models to make more accurate predictions. All forecast models will have some variances between forecasted and actual sales

* Companies typically use traditional Mean Absolute Error and Mean Absolute Percentage Error metrics to assess these forecast errors. These basic statistical measures do not show the financial impact

* Dr Catt has developed the Cost of Forecast Error (CFE) metric. This calculation includes both inventory costs and the costs of poor service (stock-outs), and incorporates the costs of holding safety stock and maintaining a desired service level

* A CFE calculation will show the trade-offs inherent in varying service level and helps determine optimal safety stocks

* Dr Catt’s research has been published by Foresight, the international journal of applied forecasting, and commented on by a number of US academics.

ENDS

© Scoop Media