Marketing Response Model

1
Marketing Response Model

• MGT453 Lecture 2

2
Agenda

• Introduce you Marketing Response Model, discuss
  its key components.
• Aggregate market response models.
• Individual response models (which can be added
  up) to represent the market.
• Calibration/Estimation.
• Criteria for selecting response models.

3
An Example (from term project)

• A manufacturer of a grocery product (e.g., peanut
  butter, ketchup, diaper) needs to find out the
  effects of marketing mix.
• It could be very hard to simply use a mental
  model to figure it out accurately.
• Marketing response models act like a tool.

4

• Marketing response models act like a tool.
• inputs the marketing actions that the marketer
  can control (e.g., price, advertising, selling
  effort, etc.), and noncontrollable variables
  (e.g., market size, competitive environment.

5
General Approach

• Inputs the marketing actions that the marketer
  can control (e.g., price, advertising, selling
  effort, etc.), and noncontrollable variables
  (e.g., market size, competitive environment.
• Response Model The linkage from those inputs to
  the measurable outputs of concerns to the firm
  (e.g., sales levels, customer awareness levels,
  product perceptions and profits).
• Objectives The measure that the firm uses for
  monitoring and evaluating those actions (e.g.,
  the level of sales in response to a promotion,
  target audience recalls an ad).

6
How to characterize a Response Model

• Number of variables.
• Include competition or not.
• Nature of relationship (linear or nonlinear).
• Static or dynamic.
• Model individual response or aggregate response.
• Level of demand analyzed (sales vs. market share).

7
An example

• Y a bX,
• where Y is sales (), X is advertising
  expenditure ().
• X is an independent variable.
• Y is a dependent variable.
• a and b are parameters.
• How to interpret a, and b?
• Say, a 23,000, b 4.

8
Aggregate response models examples

• Flexible functional form
• Y a bX cX2 dX3 eX4
• Log model
• lnY a b lnX.
• How to interpret b?
• Semi-log model
• Y a b lnX.
• How to interpret b?

9
More examples

• The exponential model
• Y a exp(bX).
• Good to characterize increasing return to scale
  (for b 0), or increasing return to decrease in
  price (for b
• The modified exponential model
• Y a1-exp(-bx) c.
• Has an upper bound (what is the upper bound?) and
  decreasing return to scale.

10
Individual Response Model

• Often times, we observe individual to make
  discrete choice decision.
• E.g., when you buy laundry detergent, you only
  choose one brand. You can only choose discrete
  number of boxes.
• The models that we have seen so far assume Y is
  continuous.

11
Calibration/Estimation

• Get parameter values so that the model is able to
  fit the observed data.
• Two main methods will be used.
• Regression
• Maximum Likelihood

12
How to evaluate a Model

• Specification
• Does the model include the right variables to
  represent the situation?
• Does the model capture the expected behavior?
• For behavior that we are uncertain, is the model
  flexible enough to allow the data to speak for
  itself?

13
Model validity

• Does the model fit the data reasonably well?
• Does the model provide value-in-use to the user?
• Goodness-of-fit is one way to differentiate
  models.