How to Incorporate Market Intelligence into Statistical Forecasting
Download as PPT, PDF9 likes5,377 views
The document discusses strategies for improving forecast accuracy at Air Products and Chemicals, Inc. through the integration of market intelligence into statistical forecasting. It outlines the company's background, the importance of forecasting segmentation, and the benefits of enhanced forecast accuracy, including reduced inventory and improved financial performance. A pilot study showed significant improvements in forecast accuracy through the effective use of market intelligence, highlighting its value in optimizing supply chain performance.
![Thank you! Stephen P. Crane, CSCP Director Supply Chain Air Products and Chemicals [email_address]](https://image.slidesharecdn.com/CDocumentsandSettingscranesteMyDocumentsSCMPresosCrane_IncorporatingMarketIntelligenceintoStatisticalForecastfinal-090222103949-phpapp02/85/How-to-Incorporate-Market-Intelligence-into-Statistical-Forecasting-39-320.jpg)
How to Incorporate Market Intelligence into Statistical Forecasting
- 1. Air Products and Chemicals, Inc. Stephen P. Crane, CSCP Director Supply Chain How to Incorporate Market Intelligence into Statistical Forecasting Orlando, Florida October 25-27 A Key to Improving Forecast Accuracy
- 2. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added Pilot Study Pilot Results Conclusions
- 3. Who is Air Products?
- 4. Air Products in Brief Global supplier of gases, chemicals, equipment, and health care services FY06 revenue ~$8.6 billion Serving customers in technology, energy, industrial and healthcare markets One of the safest large-scale chemical companies Operations in more than 30 countries
- 5. Air Products ERP Platform SAP R/3 Single Instance Initial Go-Live July 2002 (Release 1- Germany) Releases 2, 3, 4, 5 (2003 – 2006) Global Release strategy developed through 2007 APO (Advanced Planner & Optimizer) v3.0 Initial scope included deploying Demand Planning (DP) and Supply Network Planning (SNP) modules Standardized Demand Planning and Supply Planning processes for all businesses First APO forecast generated in October 2003 APO v4.1 functionality upgrade scheduled for October 2006
- 6. Global Work Processes Metrics Follow the Money!
- 7. Supply Chain KPIs Supply chain planning Forecast Accuracy % Forecast Value Added Production Plan Adherence Inventory DOS Inventory Accuracy % Master Data Accuracy (APO, production, customer) Operational Performance On-stream Efficiency Non-prime inventory Off-grade Financial Financial Forecasting Accuracy Cash-to-Cash cycle time Purchasing/Fulfillment % Orders Received on Time On-Time delivery % Perfect Order Fulfillment Customer % Complaints closed by target date Top 5 Corporate KPIs
- 8. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added FVA Pilot Study Pilot Results Conclusions
- 9. Utilizing Forecasting Segmentation Gathering Market Intelligence Reducing Forecast Variance Resulted in Improved: Inventory Accuracy Production/Supply Accuracy Order Fulfillment Performance Financial Accuracy Improving Forecast Accuracy
- 10. High Forecasting Accuracy Yields Tangible Benefits in Supply Chain Performance Companies with better demand forecast accuracy have: 15% less inventory 17% better perfect order performance 35% shorter cash-to-cash cycle times than their peers Source: AMR Research Report “The Hierarchy of Supply Chain Metrics: Diagnosing Your Supply Chain Health,” (February 2004)
- 11. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added FVA Pilot Study Pilot Results Conclusions
- 12. Demand planning needs to be based on statistical forecasting and selected market intelligence to increase the accuracy of the forecast. Forecasting segmentation should be the key analysis for prioritizing your forecasting resources. Forecasting Segmentation Forecasting Segmentation Source: Accenture High Low Statistical Forecastability (measured by 1/COV) High Sales Volume/Impact Low Rationalize/Consolidate Collaboration Rationalize SKUs, consolidate stocking locations, make to order Customer Collaboration Gather Majority of Market Intelligence Statistical Forecasting Statistical Forecasting Use statistical forecasting at an aggregate level, minimize safety stock levels Selected Account Review Q1 Q3 Q2 Q4 COV (Coefficient of Variation) = STD Deviation/Average Demand Notes
- 13. Q1 Q2 Q3 Q4 1.0 (July 2004)
- 14. Q2 Q3 1.0 (March 2005) > 98% Demand 74% Reduction Q1 Q4
- 15. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added FVA Pilot Study Pilot Results Conclusions
- 16. What is Market Intelligence? Market Intelligence is a formal process for the collection and integration of information and data about customers, products, and market segments, into an existing demand planning process, which is not typically reflected in sales order history Market Intelligence helps organizations to achieve better visibility of operational aspects of their business with improved business information resulting in optimized inventory levels and improved financial accuracy
- 17. Should you be focusing on Market Intelligence? Is your forecast accuracy as high as you would like it to be? Are you able to reliability supply products to your customers at the right place, the right time? Are you getting feedback from customers on changes in demand that you integrate into your demand planning process? Are you getting all the feedback you need from customers to properly manage their demand? Do you know which SKUs are the most important ones to focus on in managing demand?
- 18. Types of Market Intelligence Changes in customer demand which are not taken into account by statistical forecasting Future expected events Non-repeatable past events Product introduction & rationalization Sales account plans Changes in regulatory and environmental laws Changes in customer plant operations Promotions, sales, etc.
- 19. Specific Examples of Market Intelligence New or discontinued products New or lost customers Container changes Customer ship-to-location changes Unplanned customer plant outages Customer plants which are expected to be down Pre-buying, dock strikes, hurricanes, etc. Non-optimal sourcing, e.g., due to supply shortages, currency fluctuations, lack of resources, etc. Sales upside accounts
- 20. Why is Market Intelligence Needed? Forecasting process has two primary goals: Make the forecast more accurate (reduce forecast error) Make the forecast less biased Organizations fail to realize the benefits from their forecasting process for many reasons; Existing processes and measures are designed to only measure the final forecasting results Most processes fail to capture the effectiveness or “value added” by the overall forecasting effort
- 21. Why is Market Intelligence Needed? Not uncommon for many activities in a business forecasting process to fail to add value Internal politics, personal agendas, financial performance targets often skew the process, and actually make the forecast worse These non-value added activities need to be identified, improved, or just eliminated Businesses need a way to measure the effectiveness of the management judgment or Market Intelligence applied to statistical forecasts
- 22. So the fundamental problem is: Many businesses don’t know whether the efforts to apply Market Intelligence to statistical forecasts are making the forecast better or worse Many Forecast Accuracy KPIs only measure the final results of the forecasting process Forecast accuracy or variance does not tell you whether your efforts made the forecast better or worse than just using the statistical forecast alone Why is Market Intelligence Needed?
- 23. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added FVA Pilot Study Pilot Results Conclusions
- 24. Measuring Value of Market Intelligence Forecast Value Added (FVA) is designed to measure effectiveness of the Market Intelligence process Forecast Value Added (FVA) is simply the change in forecast accuracy that occurs after all adjustments have been made to the statistical forecast FVA is measured by comparing the final forecast accuracy to the statistical forecast accuracy If forecast accuracy is no better after applying Market Intelligence than without it, the Market Intelligence is not adding value and can be eliminated
- 25. Measuring Forecast Value Added By applying the measure to each material/ship-to combination: You can identify which Market Intelligence is adding value and which isn’t and from whom You can identify non-value added adjustments that can be eliminated to improve the accuracy of forecasting process FVA is designed to compare the actual demand to both the statistically generated forecast and to management’s overrides of the statistical forecast
- 26. Market Intelligence Process Gather and Submit Market Intelligence Information DAY –30 to 0 Information from Customer Contact personnel (Sales, CSO, etc) Information Received Directly From Customers Information Regarding New Products Gather Customer Data Create Demand Change Notification(s) Gather Marketing Data Information Regarding Existing Products, Markets, Segments Create Demand Change Summary Create Demand Change Notification(s) Review Demand Change Notification(s) Assess Impact of Demand Change Notifications Summarize Agreed Demand Change Notifications in Demand Change Summary
- 27. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added FVA Pilot Pilot Results Conclusions
- 28. World-Wide Leader in Vinyl Acetate Ethylene Co-Polymer Dispersion Technology, Serving Adhesives, Nonwovens, Coatings, and PSA markets Air Products Polymers, LP
- 29. Air Products Polymers Fast Facts $600 million global business Diversified geographies 65% of sales in N. America 30% Europe, 5% Asia 6 Plants (4 NA, 1 Europe, 1 Asia) 230 Products 1,800 Ship-to Customer Locations 3,500 Planning Combinations (Material-Ship-to-Primary Source Plant)
- 30. FVA Pilot Dimensions NA region selected for pilot Pilot conducted by NA Demand Manager Market Intelligence measured at the level where forecast adjustments were made FVA measured at material/ship-to level manually in APO Pilot duration 6 months, Jan-June 2006
- 31. Content Air Products in Brief Improving Forecast Accuracy Forecasting Segmentation Market Intelligence Measuring Forecast Value Added Pilot Study Pilot Results Conclusions
- 32. Pilot - FVA Results (Units in KGS) Statistical Forecast Final S&OP Forecast (Includes Market Intelligence) FVA Impact
- 33. Pilot - FVA Results Forecast Value Added Metric Measures the effectiveness of Market Intelligence The market intelligence provided by sales & marketing is having a significant impact on forecasting accuracy (Units in KGS)
- 34. Forecast Accuracy 3 Stages of Improvement Air Products Polymers, LP Data Clean-up Market Intelligence Data Aggregation World Class + 6% + 15% + 7%
- 35. Financial Forecast Accuracy 21% Improvement
- 36. Next Steps Deploy FVA KPI to Europe APO statistical forecast to be passed to BW FVA KPI to be automated using BW in Demand Planning cube FVA by Business FVA by Primary Source Plant FVA by Material/Customer Ship-to Location FVA will then be available to all business units
- 37. Conclusions Market Intelligence has a significant impact on improving forecast accuracy and effectiveness of overall forecasting process Measuring FVA from Market Intelligence provides feedback mechanism to Sales, Marketing, and customers, which improves the quality and accuracy of the Market Intelligence being collected Measuring FVA causes an organization to ask more questions about what is really going on with customers Improvement in forecast accuracy from Market Intelligence provides additional downstream supply chain benefits
- 38. Through the effective collection of Market Intelligence , a clear picture of “ what will be needed, where it will be needed, and how much will be needed” for customers, can be provided by marketing/sales organizations along with collaboration from customers and other business intelligence sources. Conclusions
- 39. Thank you! Stephen P. Crane, CSCP Director Supply Chain Air Products and Chemicals [email_address]
Editor's Notes
- #2: Introduction There’s an often-quoted saying: The only thing certain about a forecast is that it is always wrong. And yet, companies spend a great deal of time and resources trying to predict as accurately as possible the future demand of their supply chains. It is beneficial to know the future. When companies know their sales for next week, next month, and next year, they only invest in the facilities, equipment, materials, and staffing that they need. There are huge opportunities to minimize costs and maximize profits if we know what tomorrow will bring—but we don’t. Therefore we forecast. So a typical forecasting process involves historical demand data loaded into a computer database, with some form of statistical software used to generate forecasts. However, the statistical package is rarely allowed to operate on its own. Instead, a management team usually reviews and overrides the statistical forecast before giving the revised version its blessing as the official company projection of future demand. It is this process that I’ll be talking about this afternoon, how to utilize market intelligence to minimize the effort and maximize the value of the overall forecasting process.