Demand/Supply Integration: Forecasting and S&OPRemember the first day of class?Matching Supply With DemandSome Definitions So We’re All Using the Same LanguageDemand Forecasting TermsHow Do Companies Manage This Forecasting Process?The Nature of The Business Determines the Nature of The ProcessDemand Forecasting SystemsPowerPoint PresentationThere are Two Categories of Forecasting Techniques:Let’s Start with Time Series: These are the Components of Time Series AnalysisTime Series ComponentsNow let’s move on to Regression AnalysisSlide 14Slide 15Demand Forecasting TechniquesThe Other Category of Forecasting Techniques is Qualitative ForecastingQualitative Techniques And ToolsSalesforce Composite ForecastingSlide 20Demand Forecasting Performance MeasurementDimensions Of Demand Forecasting PerformanceSlide 23Slide 24(S&OP)The S&OP ProcessS&OP – It’s really a bunch of regularly scheduled meetingsS&OP Should Be Both Tactical and Strategic in NatureBy the way …Demand/Supply Integration: Forecasting and S&OP1Remember the first day of class?The right side of the room represents the Demand side of Anheuser-BuschWhat departments do you work in?What do you do?The left side of the room represents the Supply side of Anheuser-BuschWhat departments do you work in? What do you do?What do you Supply people need from the Demand people?What do you Demand people need from the Supply people?Matching Supply With Demand3SALES ANDOPERATIONSPLANNING(S&OP)SUPPLYProduction,Logistics, etc., andUpstreamSuppliersDEMANDSales and Marketing, DownstreamChannelPartnersDemand Forecast Capacity ForecastOperational PlanDemand PlanWe’ll start our discussion with forecastingSome Definitions So We’re All Using the Same LanguageDemand is what customers would buy from us if they could. This is unconstrained demand.The demand forecast is a projection into the future of expected demand, given a stated set of assumptions. This is an unconstrained forecast.Plans are the managerial actions that result from the demand/supply integration processSupply plansInventory plansDemand plans: actions Financial plans: where will we go to get the working capital that we need? 4Remember! Forecasts are best guesses about what we think will happen.Plans are decisions we make about what we will actually do.Demand Forecasting Terms5The forecasting level: At what level of granularity is your forecast expressed? (e.g. SKU, product family, etc.)•Product forecast ( how much beer will be demanded?) •Brand forecast ( how much Bud Lite will be demanded) •SKU forecast ( How much Bud Lite in 6-pack long neck bottles will be demanded)---less accurateThe forecasting time horizon: How far out into the future are you forecasting demand?Next month?Next quarter?Next 2 years?The forecasting time interval: How frequently do you update your forecast?How Do Companies Manage This Forecasting Process?6PerformanceMeasurementThe Nature of The Business Determines the Nature of The ProcessNature of the customer baseNarrow?Broad?Regional differences?Nature of the dataAge?Detail?Quality?Nature of the productsNew products?Seasonal demand?Shelf life?Project-based business?Nature of the peopleResources?Education/training?The Forecasting process that works best for one business might not work best for a different business!Demand Forecasting SystemsTwo real functions accomplished by forecasting systems:1. Statistical “engine” – software that does the statistics2. CommunicationIntegration with other corporate systems (ERP, Supply chain, etc.)Ability to share information with upstream (supplier) and downstream (customer) usersERP BackboneSystemERP BackboneSystemDataWarehouseDataWarehouseDemandForecastingSystemDemandForecastingSystemFinancialAnalysisSystemFinancialAnalysisSystemMarketing/MarketingResearchSystemMarketing/MarketingResearchSystemSupplierSystemsSupplierSystemsForecasting System OverviewCRM SystemsCRM SystemsSales Management SystemsSales Management SystemsCustomerSystemsCustomerSystemsSupply Chain SystemsSupply Chain SystemsThe important points here:1. Companies need a data warehouse to guarantee data integrity2. The forecasting system must integrate “seamlessly” with other corporate systemsThere are Two Categories of Forecasting Techniques:Quantitative, or Statistical Forecasting (which is like looking in the rear-view mirror)We’re looking for patterns in historical demandSome patterns are a function of time, and we try to identify these patterns using time series techniquesSome patterns are a function of the way that other factors affect demand, and we try to identify these patterns with regression analysis.Then, once we find these patterns, we project them into the future and voila!, we have a forecast!10Let’s Start with Time Series: These are the Components of Time Series Analysis1. Trend•Continuing pattern of demand increase or decrease•Pattern can be a straight line, or a curve2. Seasonality•Repeating pattern of demand increases or decreases•Normally think of seasonality as occurring within a single year, and cycles as occurring over longer than one year periods3. Noise•Random fluctuation•That part of demand history which the other time series components cannot explainRemember – we’re looking backward at historical demand to try and identify these patternsTime Series ComponentsWe are here, looking backward, so we can project forwardDemandTrendTrendSeasonalitySeasonalityNoiseNoiseNow let’s move on to Regression AnalysisRegression analysis is useful when you think there are measurable factors that affect demandDemand is always your Dependent Variable (Y-variable) These measurable factors are your Independent VariablesThese measurable, independent variables can be Internal or ExternalHow Regression Analysis Works – Scatter PlotDraw Regression LineRegression Statistics H H H HR Square 0.79H H H H H- CoefficientsStandard Error t Stat p-levelIntercept 716.2 467.7 1.53 0.135Ad Dollars (thousands) 22.06 1.94 11.35 0.001H H H H HDemand (thousands) = 716.2706 + 22.0580 * Ad Dollars (thousands)So, let’s say the expected advertising expenditure in August is $286,000. Our forecast of demand would be:716 + (22.06 * 286,000), or 7,025,000 unitsDemand Forecasting TechniquesRemember that these statistical forecasting techniques are looking at historical demand. With