nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]).

Do not copy or distribut

TUTORIAL

Resource Allocation
Copyright © 2018 by DecisionPro Inc.

This document is primarily intended to be used in conjunction with the Enginius

software suite. To order copies or request permission to reproduce materials, go
to http://www.enginius.biz. No part of this publication may be reproduced, stored
in a retrieval system, used in a spreadsheet, or transmitted in any form or by any
means –electronic, mechanical, photocopying, recording or otherwise– without the
permission of DecisionPro, Inc. v220322

1
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Overview
Resource Allocation is a tool used for optimizing resource size (e.g., advertising budgets) and the allocation of
resources across segments, products, channels, etc.

Hard data rarely are available to support such decisions, because resource allocation decisions influence future
(unknowable) outcomes. 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 [a resource such as advertising] 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 changes in the resources allocated.

Then, in the second phase, these calibrated response curves can be used to derive an optimal solution to the
resource allocation 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 other forms of impersonal marketing
communications? On sales promotion? On the sales force?

✓ How should individual budgets be allocated? To customers? To geographies? To sub-elements

of the marketing communications mix? Over time?

These types of questions 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 question individually.

Getting Started
The resource allocation model allows you to use your own data directly or to use a preformatted template.

Because the resource allocation model requires a specific data format, users with their own data should review the
preformatted template to become familiar with the appropriate structure.

Creating a template

The screen capture below shows the dialog box that results from using Enginius Templates (Resource allocation).

2
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

The options are as follows:

▪ General options:

1. Number of categories: A category 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 “categories” for
the purposes of this analysis.

2. Number of effort levels: The calibration of the response curve requires 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? Obtaining accurate 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.

3. Include category-specific level constraints: The number of periods to forecast will be

available if either the Advanced Bass model template or the Variable market potential
options are selected. Enter the number of periods you want to forecast. Notice that
this option simply creates placeholders for anticipated market potential and/or price
and advertising levels; it does not generate forecasts.

Note: the check box at the bottom of the dialog box will cause the template to populate with
sample (random) data that will allow you to run resource allocation immediately so you can
preview the output produced.

After selecting the desired model options, click Run to generate the data collection template.

Data blocks

The template will consist of the following data blocks depending on the template options selected.

3
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Base scenario

Data consists of current efforts and outcomes, cost, and margin for each category.

Effort data

Data consists of the levels of efforts that will be used to calibrate the response functions. “Inf” can be used to
represent an infinite level of effort.

Impact data

Data consists of the consensus estimates of how outcomes might be affected by changes in the effort variables for
each category. 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 category?

Obtaining accurate 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.

1
For a discussion and process description, please refer to the discussion of the Delphi method in Chapter 5
and Exhibit 7.9 in Principles of Marketing Engineering and Analytics, 3rd Edition.

4
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Category specific constraints (not required)

Data consists of the minimum and maximum level of effort for each category.

Running Analysis

 Each model includes a sample data set (OfficeStar) that can be

found under the Tutorials category on the Enginius dashboard.

The remainder of this tutorial uses the Resource Allocation

OfficeStar data set as the starting data set.

5
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Selecting the “Run Resource Allocation” option under the Quick Links or clicking the Resource Allocation icon on the
left will bring up the following window where you can set up your allocation scenario:

Resource allocation requires:

▪ Base scenario:

Select the data block that contains the base scenario (current levels of effort and
outcome).

▪ Response functions:

Select the data blocks that contain the effort data and impact data for setting up your
response functions.

▪ Constraints:

1. Set global minimum effort constraint: The minimum level of effort constraint. Often
this will be set to zero to prevent the software from a solution with negative levels of
effort.

2. Set global maximum effort constraint: The maximum level of effort constraint across
all categories.

3. Set category-specific effort constraints: Select the data block that contains your
category specific constraints.

6
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

▪ Run sensitivity analysis:

Sensitivity analysis performs numerous optimizations at various levels of global efforts. It

is time consuming. You may expect significant waiting time.

• Advanced

The advanced option enables an additional dropdown under the Response functions
section in which you may choose the response function model to be used for calibrating
the response curves.

1. AdBudg and Logit response functions are S-shaped functions that allow outcomes to
increase very slowly as you increase effort from low levels, and then accelerate
rapidly, before eventually reaching a plateau, or saturation level.

2. An Exponential response function increases rapidly at the very beginning, and each
additional effort unit has a marginally lower impact than the previous effort unit.

3. Automatic will select the most appropriate response function for each Category. The
response function used for each category will be indicated in the report produced.

Make the desired selections for the above data blocks and click the Run button.

 Reminder: Clicking the world icon beside the “Run” option will
allow you to choose a different output format for the report.

7
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Interpreting the Results

Response function calibration

The report will display the response functions calculated for each category.

The results are also plotted on the next chart.

8
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Unconstrained optimization

The report will provide a table and chart of the optimization process using unconstrained data compared to the
base data. (This chart will be produced even if you have placed constraints on your data.)

9
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Constrained optimization

The result of the analysis with your constraints enforced is displayed.

10
nsed to thi hong hanh nguyen, International School, Vietnam National University, Hanoi ([email protected]). Do not copy or distribut

Comparison of scenarios

A chart for each category will be produced comparing the current situation to both constrained and unconstrained
options.

11

Powered by TCPDF (www.tcpdf.org)