Our 2012 survey seeks to explore this using a survey answered by over 200 forecasters and demand planners, showing current forecasting practices and identifying gaps for future improvements, in particular in the use of judgment and statistical methods, use of internal and external information, and the use of forecasting support systems.

Past studies have addressed important issues, such as the forecasting methods used, the forecast level, frequency and horizon, use of forecasting software, or organisational responsibility and satisfaction (see, e.g., Rothe 1978, Sanders & Manrodt 1994, Winkelhofer & Diamantopoulos 1996, Moon & Mentzer 2003). However, much time has passed since their original surveys. With more data becoming available, from ePOS-sales to inventory in distribution centres, and at increasing granularities from monthly to weekly, daily or even intraday data. ERP and Advanced Planning, and forecasting support systems such as SAP APO DP are maturing to handle such data.

At the same time, new organisational processes have developed, firmly establishing information sharing and collaboration (ISC) schemes in business practice. These take the form of ISC between departments within a company (i.e. Sales & Operations Planning, S&OP) as well as between companies along the supply chain (i.e. Vendor Managed Inventory, VMI, and/or Collaborative Planning Forecasting and Replenishment, CPFR). While these developments are all widely reported to deliver improved performance and profitability, it raises the question if forecasting practices have changed as well?

In contrast to earlier studies, our survey pays particular attention to the challenges of forecasting in today's collaboration-rich environment, the resulting information overload and how leading companies are addressing it. These issues had not been subject to empirical analysis yet, raising a number of important questions: What data is being exchanged? In what form, how often, and supported by which systems? And ultimately: how is this data being used in (better?) forecasting.

Selected key findings are as follows:

• CIS partnerships are prevalent but inconsistent. Manufacturers are heavily involved in internal (S&OP) and external (VMI, CPFR) collaborative schemes, typically running multiple schemes in parallel.
• Downstream data of various types (e.g. forecasts, POS data, stock levels, promotions) is often available, typically shared through unstructured or semi-structured methods (e.g. email, phone). Information exchange across different trading partners is not reported to be consistent.
• Forecasting typically relies on judgement and basic time series methods.
• In terms of their overall approach, firms rely on some degree of judgement in most of their final forecasts.
• Most companies use available external data in forecasting through a mixture of statistical and judgemental means. A greater amount of judgement is used on the external data than with internal data.
• The statistical methods used are dominated by simple univariate methods such as smoothing, averages, intermittent and naive (85%).
• Advanced methods able to deal with additional variables (e.g. promotions, product introductions, EPOS data) are seldom used (15%).
• Manufacturers almost always forecast in monthly buckets, only sometimes also forecasting in weekly and occasionally daily buckets.

(Note: from the sample through our mailing list of past collaborators and LinkedIn groups, the responses may be biased towards (a) companies more active in forecasting development and hence higher forecasting maturity in methods, systems and processes, and (b) companies more engaged in information exchange and collaboration schemes such as S&OP, VMI and CPFR).

This short  summary only provides an overview of few key findings. The complete report (65 pages, analysed by collaboration type) is available from the authors upon request. Please contact Matt Weller, PhD candidate at the Centre for Marketing Analytics and Forecasting, or Dr Sven F Crone, Director of the Lancaster Centre for Forecasting, to obtain a full copy.