RESSOURCE ALLOCATION TUTORIAL 1/16 MARKETING ENGINEERING FOR EXCEL  TUTORIAL  VERSION 1.0.11 Tutorial Resource Allocation Marketing Engineering for Excel is a Microsoft Excel add-in. The software runs from within Microsoft Excel and only with data contained in an Excel spreadsheet. After installing the software, simply open Microsoft Excel. A new menu appears, called “MEXL.” This tutorial refers to the “MEXL/Resource Allocation” submenu. Overview Resource allocation analysis is a tool used to optimize resource sizing (e.g., advertising budgets) and resource allocations (across segments, products, channels, etc.). Hard data rarely are available to support such decisions. Consequently, in the first phase, resource allocation analysis builds on managerial experience and insights to create an effort/impact response curve consensus; that is, users answer the question, “Given our experience and knowledge about the market, products, customers, and competition, what would happen if we increased […] by x%?” A response curve may then be calibrated on the basis of these “what-if” assessments to determine how the market might react to various changes. Then, in the second phase, these calibrated response curves suggest an optimal solution to the problem at hand by taking into account both stated objectives and constraints (e.g., budget limitations). Resource allocation analysis helps firms answer such questions as:  How much should we spend in total during a given planning horizon?  How should that spending get allocated to each marketing mix element? How much of our budget should be spent on advertising and RESSOURCE ALLOCATION TUTORIAL 2/16 other forms of impersonal marketing communications? On sales promotion? On the sales force?  How should individual budgets be allocated? To customers? To geographies? To subelements of the marketing communications mix? Over time? These three question areas are closely interrelated. It is nearly impossible to address the question of how much to spend (budget) without determining how to spend the budget properly (i.e., allocated across competing uses). Thus, these questions provide the perspective used to explore each element individually. Getting Started To apply resource allocation analysis, you can use your own data directly or a preformatted template. However, because of the required calibration phase, we urge users to follow the steps described next. The next chapter explains how to create an easy-to-use template to enter your own data. If you want to run a resource allocation analysis immediately, open the example file “OfficeStar Data (Resource Allocation, calibrated).xls” and jump to “Step 4: Running analyses” (p. 10). Running analyses require calibrating response curves based on managerial judgments. If you want to perform that intermediary step yourself, open the example file “OfficeStar Data (Resource Allocation).xls”, which contains managerial judgments on which to calibrate the curves, and jump to “Step 3: Calibrating the response curves”. By default, the example files install in “My Documents/My Marketing Engineering/.” RESSOURCE ALLOCATION TUTORIAL 3/16 Step 1 Creating a template Using the interactive assistant In Excel, if you click on MEXL  RESOURCE ALLOCATION  CREATE TEMPLATE, a dialog box appears. This box represents the first step in creating a template for running the resource allocation software (both calibration and optimization phases). The first dialog box prompts you to use an interactive assistant. Unless you are already familiar with the methodology, you should select “yes.” Specifying the measurement and calibration base The first step of the template generation process requires you to label the measurement and calibration units.  Effort unit describes the “input” label of the model, usually the decision variables on which you can act, such as how many advertising dollars you want to invest to certain channels or how many sales representatives you plan to allocate to a specific product, market, or segment of customers.  Outcome unit describes the “output” label of the model, including the consequences of the decisions you have made, such as sales, revenues, profits, unit sold, market shares, or generated leads.  Calibration can be input in either absolute or relative terms. For instance, if 50 sales reps generate $1M in sales, you could calibrate the response curve by saying, “If we hired 25 additional sales reps, by how many dollars would our sales increase?” or “If we hired 50% more sales reps, by how many percentage points would our sales increase?” or use a mix of both questions. You should use whatever measures are most convenient and natural for you. (Note: If you start from nothing, use absolute measures.) RESSOURCE ALLOCATION TUTORIAL 4/16 Listing segments The second step of the template generation process in the interactive assistant requires you to input the segment names. A segment is the base unit of analysis in resource allocation; you might allocate resources across channels, customers, or products, such that each channel, customer, or product is defined by the generic term “segment” for the purposes of this analysis. RESSOURCE ALLOCATION TUTORIAL 5/16 Specifying current situation The third and last step of the template generation process requires you to describe the current situation, in terms of current efforts, outcomes, and cost and gross margin figures. When you click “Next >,” a new spreadsheet will be generated that contains a “Scenario” sheet and a “Calibration” sheet (see the next section for a snapshot of the latter). RESSOURCE ALLOCATION TUTORIAL 6/16 Not using the interactive assistant If you choose not to use the interactive assistant, a unique dialog box will appear, requesting you to specify the number of segments and the effort and outcome calibration bases. When you click “OK,” you generate a new template. You must enter segment labels, current situation, cost and gross margin figures, and other inputs manually. Step 2 Entering your calibration data In this tutorial, we use the example file “OfficeStar Data (Resource Allocation).xls,” which in the default condition appears in “My Documents/My Marketing Engineering/.” To view a proper data format, open that spreadsheet in Excel. A snapshot is reproduced below. RESSOURCE ALLOCATION TUTORIAL 7/16 The calibration sheet contains the consensus estimates of how outcomes might be affected by changes in the input variables for each segment. For instance, how much additional sales would be generated by an increase of 50% in the sales force allocated to a specific customer segment or product? To obtain proper and consistent consensus estimates from a group of managers and experts is as much an art as it is a science. It requires free information exchange, constructive feedback, and a constructive discussion of the “what-if?” scenarios.1 Step 3 Calibrating the response curves After you enter consensus estimates for different “what-if” scenarios for each segment, click on MEXL  RESOURCE ALLOCATION  CALIBRATE RESPONSE CURVES. The following dialog box appears: Options The first option refers to generating a diagnostic workbook, which gives you details about the estimated parameters and goodness of fit, as well as a graphical representation of the fitted curves. You can also specify the functional form of the response curves you want and thereby fit your data:  AdBudg and Logit functions are S-shaped functions that allow outcomes to increase very slowly as minimal efforts increase, then rapidly accelerate to eventually reach a plateau, or saturation level.  An Exponential function increases rapidly at the very beginning, and each additional effort unit has a marginally lower impact than the previous effort unit. 1 For a discussion and process description, please refer to Principles of Marketing Engineering, 1st ed., Exhibit 7.3, p. 162. RESSOURCE ALLOCATION TUTORIAL 8/16 The AdBudg function is the default selection, because it is the most versatile and recommended response curve functional form. If you want functional forms to differ among segments or prefer to calibrate one response function at a time, you can do so. Simply proceed through the described procedure and then later select only those cells relevant to your segment of choice. You can select one, some, or all segments (the latter is the default option). The segments simply need to be in contiguous cells. Selecting calibration data, base scenario, and target cells The next step asks you to select calibration data, base scenario data, and the target cells (recommended scenario data) to calibrate the response curves. If you used the “Generate Template” option of Marketing Engineering for Excel, these cells will be selected by default. Diagnostic workbook If you have selected “Generate a Diagnostic Workbook,” a new spreadsheet will be generated, showing how the calibrated curve fits the calibration data entered by the managerial team. It also displays example points and response function–estimated parameters. Notice that, by design, the parameters will be estimated such that the response curve replicates the lower and higher bounds, as well as the current situation, as closely as possible, even if that means fitting the other points more poorly. RESSOURCE ALLOCATION TUTORIAL 9/16 Response curve estimations After calibrating the curves (regardless of whether you have generated a Diagnostic Workbook), the calibrated response curves get copied into the initial spreadsheet, in the selected Recommended Scenario section. Although at first you will see no difference between the base and recommended scenarios, the key difference is that the outcome in the recommended scenario section is not a given anymore; rather, it is a function of the current efforts. If you alter the current efforts in the recommended scenario section, the estimated outcomes will change according to the calibrated response curves. For instance, in the OfficeStar example, suppose the user cuts the advertising budget for television ads by half and double the budget for print ads. On the basis of the calibrated response curves, the model estimates that total net margins will diminish by more than half a million dollars (from $5.8M to $5.2M). RESSOURCE ALLOCATION TUTORIAL 10/16 After you have generated this spreadsheet, you can “play around” and see how an increase or decrease of effort in one segment affects the overall results. Marketing Engineering for Excel, however, offers you a better and more straightforward way to find the optimal budget and budget allocation, given specific objectives and constraints. This method is the point of the next section. Step 4 Running analyses When the spreadsheet is ready for optimization, click on MEXL  RESOURCE ALLOCATION  RUN ANALYSIS. The following dialog box appears: RESSOURCE ALLOCATION TUTORIAL 11/16 This option enables you to find the optimal effort levels (e.g., total budget, budget allocation across segments) to maximize outcomes (e.g., net revenues), conditional on specified constraints. Options The dialog box offers you several options from which to choose. The first option enables you to specify constraints on your optimization, that is, lower and upper bounds within which the solution must fit. If you click on this option, another dialog box will appear later (see below). The second option, if checked, generates a Diagnostic Workbook that details the recommended solution. This workbook may contain the following information:  Base vs. Recommended Scenario Chart: A simple, graphical comparison of the recommended solution and the current situation.  Opportunity Costs: The extent to which imposed constraints limit the final results, formally defined by the additional net margin achieved by releasing a given constraint by 1 unit, per segment or in total. For example, if you impose a constraint stating that there can be no more than 50 sales representatives allocated to a given line of product or that the total advertising budget cannot be more than $1.5M, this function indicates how much the bottom line would improve if the constraints were relaxed by one unit (i.e., 51 sales reps, $1.6M). The literature sometimes refers to the term shadow price, formally defined as the additional net margin achieved by increasing a given constraint by 1 unit. Shadow prices and opportunity costs are closely related in the following ways: When the user imposes an upper-bound constraint (say a segment cannot have more than 50 sales reps, i.e., X  50), shadow prices and opportunity costs are equivalent. When the user imposes a lower-bound constraint (say a segment cannot have less than 20 sales reps, i.e., X  20), on the one hand, shadow prices will compute differences in net margins with a new constraint of X  21, and that figure might actually be negative; you may loose additional money by increasing an already costly constraint. On the other hand, opportunity costs will compute differences in net margins with a new constraint of X  19, a figure that by definition cannot be negative; you cannot waste additional money by relaxing a constraint. At best, it doesn’t affect anything in your strategy.  Sensitivity Analysis: An assessment of maximum potential. It shows the maximum profit that could be achieved by imposing various levels of constraints on the total efforts (e.g., advertising budget, size of the sales force), given optimized resource allocation. Selecting base scenario, recommended scenario, and target net margin cells You next will be asked to select different range cells to feed the optimization algorithm. If you have used the “Generate Template” option of Marketing Engineering for Excel, these cells are selected by default. RESSOURCE ALLOCATION TUTORIAL 12/16 Imposing constraints If you checked the “Impose Constraints” checkbox, a new dialog box appears, with which you can impose lower and/or upper constraints at the global level (e.g., a maximum allowed advertising budget) or the level of each segment (e.g., a minimum level of spending for a given channel). RESSOURCE ALLOCATION TUTORIAL 13/16 By default, the lower-bound constraints are imposed, with a minimum level of 0. Most problems have lower-bound constraints. For instance, you cannot spend less than nothing in a given segment. Therefore, you must impose lower-bound constraints of 0 (the default constraint); otherwise, the software might recommend a nonsensical solution. Step 5 Interpreting the results Base vs. recommended scenario The recommended scenario (efforts) moves into the initial spreadsheet. The copy generated in a Diagnostic Workbook (if you have checked this option) creates the following chart: In the OfficeStar example, the Direct Marketing budget should be almost doubled, because this channel offers the most promising upstream potential. The budget for radio ads, in contrast, seems to have reached a satisfactory level. On the basis of its calibration and analysis of the management team’s input, the model suggests that the company is grossly underinvesting in advertising and promotion and that an increase and better allocation of its total advertising and promotional budget (from $3.15M to $5.79M) could almost double gross revenues and increase total net margins from $5.8M to $10.7M. RESSOURCE ALLOCATION TUTORIAL 14/16 Opportunity costs Opportunity costs represent the additional net margins a company could attain by releasing a given constraint by 1 unit (either decreasing a lower-bound constraint, or increasing an upper-bound constraint), per segment or in total. In the OfficeStar example, we do not impose any constraints; hence, none could be released to improve total net margins. If we rerun the analysis and constrain direct marketing spending to a maximum level of $2M (2,000 units as opposed to the recommended 3,351), perhaps because management does not want to invest too heavily in that channel, the predicted total net margins will be much lower, because all revenues potentially generated by direct marketing get constrained in turn. RESSOURCE ALLOCATION TUTORIAL 15/16 In this case, the software reports an opportunity costs for direct marketing of $3,771. At that level of spending, releasing the constraint by one unit (i.e., increasing the maximum direct marketing budget by $1,000) increases total net margins by $3,771. Therefore, the return on investment of an additional dollar spent in direct marketing is +377%. Sensitivity analysis The sensitivity analysis feature graphs the evolution of net profit (in thousands of dollars) while the total effort varies. There are no constraints on segment- level marketing resources in the sensitivity analysis. Several specific data points are used to generate the graph:  No effort  The optimal solution  The optimal solution +50% effort (over-investing)  Various datapoints in between This chart gives a general overview of the sensitivity of the total net profit to major changes in the level of selling effort. RESSOURCE ALLOCATION TUTORIAL 16/16