MARKETING 5600/6600
Section 1
Draft
Marketing Analysis and Decision Making in an Information Age
Spring 2007
|
Bill Moore |
MW 9:
|
|
Office: KDGB 107 |
CRCC 210 |
|
Phone: 581-5023 (office) |
|
|
|
|
|
E-Mail
mktbm@business.utah.edu |
|
COURSE OBJECTIVES
This course deals with concepts, methods, and applications of decision modeling to address marketing issues such as segmentation, targeting and positioning; product design; advertising, salesforce, and promotion budgeting; and pricing. It will attempt to translate conceptual understanding into specific operational models that can be implemented with PC-based software. Specifically, the course objectives are to:
· Provide you with an understanding of the role that analytical techniques and computer models can play in enhancing marketing decision-making.
· Improve your skill in viewing marketing processes and relationships systematically and analytically.
· Expose you to examples that demonstrate the value of the analytical approach to marketing decision-making.
· Provide you with the software tools that will enable you to apply these models to real marketing problems.
REQUIRED COURSE
MATERIAL
Gary L. Lilien and Arvind Rangaswamy (2004), Marketing Engineering: Computer Assisted Marketing Analysis and Planning, Revised 2nd Edition, Trafford Press.
This book can be purchased at Amazon.com. Alternatively, there are about 15 copies of the book at the book store that can be purchased for a slightly lower price than on Amazon. Make sure you get a new copy of the book.
There is some material to help you install the software at: http://www.mktgeng.com/support/tutorials.html then click “Download Installation Manual”
GRADING
The grades will be weighted as follows:
|
|
Written Cases Class
Participation Quizzes First Conjoint Analysis Project Second Conjoint Analysis Project MDS and Clustering
Project ADBUDG
Spreadsheet Project |
40% 20% 20% 5% 5% 5% 5% |
Class participation is also an important
indication of daily preparedness.
Thoughtful participation facilitates the learning process and makes classes
lively and interesting. Therefore, your
grade for class participation is based on the quality of your discussion
during the cases and exercises. Absences, for any reason, will result in
a zero in class participation for any case discussion missed. A maximum of two absences can be made up by
handing in a written case or homework exercise within one week of the
missed session or during the last class meeting, which ever comes first.
I will attempt to distribute air time evenly, but
that is a difficult job. Class
participation is both of our responsibilities.
If you feel that you are not given an adequate chance to participate,
come see me early in the term. If you
are particularly upset during a discussion, stop the class and say I am not
calling on you. I will not be
offended. Do not wait until after final
grades have been handed out before voicing a complaint.
You may not always be well prepared for a
specific case due to personal circumstances.
Although you will not know in advance if your name is on my call list
for that day, it is a good idea to inform me before class if you are
unprepared and wish to postpone formal participation. You do not need to inform me as to why
you wish your name removed from my list, if it is on my list. Obviously, I do not recommend these requests
as a repeated strategy! On the other
hand, if you are not prepared, and you take your chances that your name will
not be called, you run the risk of zero participation points for one of your
few formal participation opportunities.
I will have a list of alternative participants; I will make a last
minute substitution if you are on my list and request to be removed.
Two cases must be submitted in written form and
must be prepared in groups of two or three (exceptions must be cleared with
me). These will be assigned approximately ten days before they are
due. They are due in class on the date the case is discussed and will
not be accepted late.
These word-processed reports should be in the
form of a business memo to a person in the case and cannot exceed 1000
words (approximately 4 pages). Tables and exhibits, which are not subject
to this 1000-word limitation, should be placed in an appendix. Leave
1" margins for my comments. Write your name only on the
backside of the last sheet.
There will be approximately 10 to 12 quizzes during the term. You will be given no more than 10 minutes to answer each, so you will need to bring the answers to class rather than try to calculate them during the quiz. If you do not bring a laptop to class, make sure you print out your analysis (which you should do for the discussion anyway). Quizzes will generally cover numbers, like parameter estimates or optimal levels of marketing effort, that will be the start of our classroom discussions.
In addition to the academic work described above,
students are required to spend about an hour participating in a consumer
research study. This task is viewed as a
course requirement and may contribute toward the student's understanding of
data collection and manipulation techniques discussed in the course.
You can access overheads for class, spreadsheet
and data files, and various instructional aids at the class website http://www.business.utah.edu\~mktbm\mkt6600. Communication will be via WebCT.
The University of
Utah David Eccles School of Business seeks to provide equal access to its
programs, services and activities for people with disabilities. If you will need accommodations in this
class, reasonable prior notice needs to be given to the instructor and to the
Center for Disability Services, http://disability.utah.edu/,
160 Olpin Union Building, 581-5020 (V/TDD) to make arrangements for
accommodations. All written information
in this course can be made available in alternative format with prior
notification to the Center for Disability Services.
Class
Schedule
|
Date |
Topics & Assignments |
|
Introduction · Marketing decision making · Using models to improve decision making · Linear response models · Regression with Excel |
|
|
Chapter 1, Chapter 2 (pp. 27-32 linear response models) Marketing Engineering Notes: Response Models – Linear Regression http://www.business.utah.edu/~mktbm/mkt6600/MarketingEngineeringNotes.doc |
|
|
1. Open the Allegro.xls file. First read the instructions on the Intro Worksheet. Then use the Allegro “dumb” spreadsheet, Sheet1, to decide how to improve next year’s performance. Then repeat with the “smart” spreadsheet, Sheet2. http://www.business.utah.edu/~mktbm/mkt6600/ALLEGRO.XLS |
|
|
Wed Jan 10 |
Conjoint Analysis I |
|
Handout Conjoint Analysis http://www.business.utah.edu/~mktbm/mkt6600/CJA.doc |
|
|
Analyze your own conjoint data. |
|
|
Wed Jan 17 Buc 103 |
Conjoint Analysis II |
|
Chapter 7 (pp. 233 – 52) |
|
|
Explore the Conjoint
Analysis software (p. 272-6) C:\Program Files\MktgEng\tutorials\conjoint.pdf 1. What are the average conjoint weights among the 40 people who filled out the questionnaire? 2. Describe the four existing hotels. 3. What is the estimated market share of the four hotel designs already selected? (Use the max utility choice rule for all market share estimates.) 4. Be prepared to set up a conjoint analysis on a product or service of your choosing. Think of two or three attributes and two or three levels on each. |
|
|
First Conjoint Analysis Project |
|
|
Mon Jan 22 |
Conjoint Analysis –
Application |
|
Forte Hotel Case (p. 272-6) |
|
Questions:
|
In the conjoint software, the last two options under analysis are “choose optimal products from a specified set,” which I will call the “simulator” as it simulates the share of any given new hotel and “view optimal products,” which I will call the optimizer as it gives us the “right answer” the hotel with the highest market share. 1. Assume you do not have the simulator or the optimizer with this software. So you have information on the individual conjoint weights, the average weights, which you can get from the bottom of the screen of the “generate conjoint matrix” option, and the average weights from each of the three benefit segments from the cluster analysis program. You also know the design of the competitive hotels from the “create/edit existing product profiles” option of the analysis menu. Given that information and your own judgment how would you choose the optimal hotel design? Put that design(s) in the simulator (by adding them through the create/edit new product profiles option under analysis) and see what market share it gets. 2. Look at the four designs that Hotel Forte management has come up with, Professional1, Professional2, Tourist, and Deluxe. Based on what you know of customer preferences (the conjoint weights) and the existing hotels, what do you think of them? What changes would you suggest? Try the changes with the simulator. 3. Examine the four “optimal” profiles. What do you think of them? Which would you choose? 4. What other analysis would you like to be able to perform? |
Wed Jan 24 |
Segmenting & Targeting · How can market segments be identified and distinguished? · Which segments offer the highest potential? |
|
Chapter 3 (pp. 61-96) |
|
|
Mon Jan 29 Buc 103 |
Segmentation and Targeting cont’d · Conducting segmentation studies |
|
Marketing Engineering Notes - Cluster Analysis http://www.business.utah.edu/~mktbm/mkt6600/MarketingEngineeringNotes.doc |
|
|
Explore the Cluster Analysis Software C:\Program Files\MktgEng\tutorials\cluster.pdf 1. Cluster the following data set: http://www.business.utah.edu/~mktbm/mkt6600/pda.dat a. Determine the appropriate number of clusters b. Describe them 2. Discriminate with the following data set: http://www.business.utah.edu/~mktbm/mkt6600/pda_dis.dat a. Describe each of the segments demographically Cluster the pda.dat to develop a 3, 4, 5, and 6 cluster solution use the pda_dis.dat for discrimination of each of those numbers of clusters (Put these solutions in a spreadsheet like we did in class for ease of comparison), identify the difference between succeeding solutions (i.e., how does the four cluster solution differ from the three cluster solution?), determine the appropriate number of clusters to use as the basis of your analysis, and think about which segment you would direct your marketing efforts to. Look at the PDA2001 Questionnaire on page 107. The questions X1 to X8, X14 and X15, and Z1 to Z17 are the same as the data in pda.dat and pda_dis.dat. b. |
|
|
Wed Jan 31 |
Segmentation – Application I |
|
Conglomerate Inc.’s New PDA (2001) Case (pp. 104 – 112) Modifications to the questions in the book: 1. How many segments did you use as the basis of your analysis?
2. What segment(s) did you choose to target? a. How would you describe that segment in terms of both needs and demographics?
3. Describe your marketing mix for attacking those segments. 4. What concerns do you have with this approach? 5. What are the next steps you would recommend for Netlink and the development of ConneCtor? |
|
|
Mon Feb 5 Buc 103 |
Second Conjoint Analysis Project |
|
Assignment: |
Re-examine the Conjoint
Analysis software (p. 272-6) C:\Program Files\MktgEng\tutorials\conjoint.pdf 1. Start to design a simple conjoint analysis questionnaire by determining the attributes and levels for a product or service. 2. Input this into the conjoint analysis software and save. 3. Design two or three competitive products or services. |
|
Wed Feb 7 |
Conjoint Analysis and Clustering Application |
|
|
Dürr Environmental Inc. Air Pollution Control Systems 1. What products should you offer, which segments should you target, and what should your selling proposition be? 2. What is the forecast marginal revenue of your recommendation versus other possible recommendations? |
|
Mon Feb 12 |
Multinomial Logit Models |
|
Chapter 2 (pp. 43 –50), Chapter 3 (pp. 96 – 103) |
|
|
1. Explore the Multinomial Logit Software C:\Program Files\MktgEng\tutorials\mnl.pdf a. Estimate the transportation response functions using both regression http://home.business.utah.edu/~mktbm/mkt6600/TranschReg.xls and multinomial logit models http://home.business.utah.edu/~mktbm/mkt6600/transch.dat . a. Estimate the response MASSMART functions using both regression http://home.business.utah.edu/~mktbm/mkt6600/MassMartReg.xls and multinomial logit models http://home.business.utah.edu/~mktbm/mkt6600/MassMartChoice.dat . |
|
|
Wed Feb 14 Buc 103 |
Logit Models |
|
Assignment |
2. Explore the Multinomial Logit Software C:\Program Files\MktgEng\tutorials\mnl.pdf a. Estimate the response MASSMART functions using both regression http://home.business.utah.edu/~mktbm/mkt6600/MassMartReg.xls and multinomial logit models http://home.business.utah.edu/~mktbm/mkt6600/MassMartChoice.dat . b. Estimate the ABB logit model using the abblogit.dat file. 3. Explore the Choice-Based Segmentation Software (pp.113 – 6) C:\Program Files\MktgEng\tutorials\abb.pdf |
|
Second Conjoint Analysis Project |
|
|
Wed Feb 21 |
Segmentation and Targeting – Application II |
|
ABB Case (pp.113 – 6) Modifications and clarifications to the questions in the book: 1. In question 3, an equivalent assumption (regarding incremental revenue) is that if you do not target a firm in the Competitive Segment for extra effort, your chance of getting that customer is 50%, but if you target it, your chance of getting the customer is 100%. If you do not target a firm in the Switchable Segment, your chances of getting that customer is 0%, but if you do target that customer, your chance of getting it is 50%. In both cases, targeting a customer in either of these segments increases your chance of getting it by 50%. 2. After answering question one, assume that the assumptions in the model are correct, i.e., you will gain 50% of the Switchable and Competitive customers you target but your efforts will have no incremental impact on the Loyal and Lost Segments, what will be the increase in your revenue? 3. For question 3, estimate how much revenue will increase under these assumptions if you target the best customers according to the model. |
|
|
Mon Feb 26 Buc 103 |
Segmentation and Targeting – Application III · Targeting segments of one · Data mining and artificial neural networks |
|
Chapter 5 (pp. 170 – 173) |
|
|
Assignment: |
Start the analysis of the Bookbinders Book Club case 1. Estimate a response model using regression on the BBBC.xls file 2. Estimate a logit response model on the BBBC.dat file 3. Determine which customers in the BBBCPred.dat file you would focus you effort on if you could target 10% of them. |
|
Wed Feb 28 |
Targeting – Application
III |
|
The Bookbinders Book Club (pp. 185 - 88) |
|
|
Assignment: |
1. Use both a regression and logit model for the following: a. Estimate the relationship between the independent variables and response to the offer on the 1600 people in bbbc.dat or bbbc.xls. b. Use those coefficients to predict the likelihood of response from the 2,300 people in the bbbcpred.xls file. c. Order those customers from best to worst prospect. d. Use their actual responses to determine how many and what percent of the people would have responded if you would have taken the best potential prospects (e.g., the 10% best, 20% best, etc.) e. Choose the percentile of potentially best prospects that would maximize your profits. 2. Answer questions associated with the case. |
|
Mon March 5 |
Product Positioning ·
How do consumers perceive the brands in a market? ·
How can a product differentiate itself on a key perceptual dimension? |
|
Chapter 4 (pp. 117 - 46) |
|
|
Wed March 7 Buc 103 |
Product Positioning – Application |
|
Explore the Product Positioning Software and the Infiniti G20 Case (pp. 148 - 54) C:\Program Files\MktgEng\tutorials\positioning.pdf 1. Do a positioning analysis using the class' perceptions and preferences. |
|
|
Mon March 12 |
Product Positioning – Application |
|
Infiniti G20 Case (pp. 148 - 54) For all questions use both the tabular and perceptual mapping data. 1. Describe the underlying perceptual dimensions of the space. If appropriate, use multiple attributes to describe each dimension. Which dimensions are easiest (or hardest) to interpret? How much variance is explained by each? 2. How do people perceive the Infiniti? How well is it positioned? The Infiniti was promoted as a Japanese car with a German feel. What do you think of when you hear “Japanese car” and “German feel”? Is this a credible claim given the perceptions and preferences of the respondents? 3. What attributes are most important to each segment? Which cars are most preferred by each segment? 4. To which segment(s) would you market the Infiniti? How would you reposition it? First where would you try to move it in the space? Second, what attributes would you change to move it to that position? Briefly describe your marketing program. 5. What ongoing marketing research program would you recommend to Infiniti to improve its performance? What do you recommend that Infiniti do differently in terms of marketing research? |
|
|
Wed Mar 14 |
Summary Lecture and Introduction to Linear Response Models |
|
MDS and Clustering Project |
|
|
Chapter 2 (pp. 27-32 linear response models) Marketing Engineering Notes: Response Models – Linear Regression http://www.business.utah.edu/~mktbm/mkt6600/MarketingEngineeringNotes.doc |
|
|
Assignment: |
1. Use the Medical Advertising spreadsheet to determine the responsiveness to weight loss advertising. First read instructions on Sheet0, then look at Sheet0.5. The other sheets and charts have my solutions. http://www.business.utah.edu/~mktbm/mkt6600/MedAdv.xls |
|
Mon – Fri March 19-24 |
Spring Break |
|
Mon Mar 26 |
Using Solver to estimate nonlinear response models ·
ADBUDG ·
Multiplicative models ·
Elasticity |
|
Chapter 2 (pp. 33- 43 and appendix on Excel’s Solver) Marketing Engineering Notes: Linearizable Response Models, Estimating Nonlinear Models with Solver, and ADBUDG http://www.business.utah.edu/~mktbm/mkt6600/MarketingEngineeringNotes.doc |
|
|
1. Complete the exercise on advertising budgeting with a linear response function – Set medical information budgets for two of the four remaining categories. Last names A – L do Quit Smoking and Cancer and last names M – Z do Physician Referral and Medical Information. 2. Study the example of estimating nonlinear least squares functions in: http://www.business.utah.edu/~mktbm/mkt6600/NonlinearLeastSquares.xls Perform a non linear least squares estimation for one of the other medical response functions you estimated earlier. You should get the same answer. 3. Study the response functions for data in the Nonlinear Advertising spreadsheet. http://www.business.utah.edu/~mktbm/mkt6600/NonlinearAdvSales.xls Choose the most appropriate response function. |
|
|
Wed March 28 |
Practice using Solver to estimate non linear response functions · ADBUDG · Profit Functions |
|
Marketing Engineering Notes: ADBUDG |
|
|
1. Estimate linear, ADBUDG, and multiplicative response functions with data in the example on p. 38. For multiplicative function, estimate the model only over the last seven observations. You may use either the Response Modeler C:\Program Files\MktgEng\tutorials\modeler.pdf or Excel to estimate the nonlinear functions. (I strongly recommend Excel.) I get different answers than in the book. For example, a = 36.796 and b = 4.734. 2. After you have estimated the ADBUDG function, assume that price is $1.00 and variable cost is $.25. Graph the profit function over a range of values of marketing effort between 0 and 30, and determine the optimal level of marketing effort. Using both: a. A graphical method to approximate the optimal level b. Solver to find the exact optimum 3. Study the way to estimate the ADBUDG function from judgmental data. This will be used in the Conglomerate Promotional Analysis next week. See the spreadsheet: http://www.business.utah.edu/~mktbm/mkt6600/ADBUDGJudgment.xls |
|
|
Mon April 2 Buc 103 |
Response Models – Applications |
|
1. Explore the Conglomerate Promotional Analysis Software (pp.58 - 60) C:\Program Files\MktgEng\tutorials\conglom.pdf a. Fill
in formulas for gross and net profit for b. Determine
the optimal promotional level for c. Use
0% as a starting value and find the optimal promotional level for d. Recalibrate
the response function for e. Look at the discussion questions for November 3. Do you have any questions about them? |
|
|
Wed April 4 |
Response Models – Application |
|
Conglomerate, Inc. Division Exercise (pp.58 - 60) 1. Do the analysis suggested in the case for the four scenarios on pages 58 - 59. a. In scenario 2, the TOTAL promotional budget should be less than or equal to the planned TOTAL budget, $2,013,437, but is reallocated across the four cities. b. In scenario 3, set three of the four cells equal to the remaining one, but allow ALL four cells to vary. 2. Estimate the ADBUDG parameters using the data on page 60. Then build a profit function, graph it, and use solver to find an optimal level of promotional spending. 3. What happens to the optimal spending level if the market saturates at 13.8M units rather than 11.8M units? What happens to the optimal spending level if the cost of delivery drops to $1.39 fro $1.69 per unit (use the old saturation level)? |
|
|
Mon April 9 |
Advertising Budgeting and Sales Force Sizing and Allocation Sales force Management ·
Sales force decisions · Sizing and allocation of sales force efforts ·
Call efficiency |
|
Chapter 8 (pp. 302-19) Chapter 9 (pp. 354 – 66, 373 - 79) |
|
|
1. Use the Allegro Case spreadsheet, http://www.business.utah.edu/~mktbm/mkt6600/AllegroCase.xls to estimate a multiplicative model using both (1) the market place data and a regression and (2) the judgmental data to build a response model and smart spreadsheet. Find the optimal level of price and advertising for each model.
|
|
|
Wed April 11 Buc 103 |
Advertising Budgeting
and Sales Force Sizing and Allocation cont’d |
|
Explore the Sales Resource Allocation Software (pp. 386 - 408) C:\Program Files\MktgEng\tutorials\reallocator.pdf |
|
|
Spreadsheet Project |
|
|
Mon April 16 |
Sales Force Allocation – Application |
|
Syntex Laboratories (A) Case (p. 386 - 408) As you are combining the model with your judgment to answer the following questions, think about: 1. At what levels of budget and allocation do you have the most confidence in the model’s predictions? 2. If you feel Syntex is misallocation effort to one product or specialty, what do you think about its allocations to other products or specialties? 3. The model is based on a number of assumptions, with which ones are you most and least comfortable? |
|
|
1. If the sales force is maintained at its current size (about 430 reps), how does the current allocation compare to the optimal allocation as indicated by the model? 2. What sales force size and allocation would you recommend for each of the next two or three years? Would you allocate sales calls by product, specialty, or some combination? 3. What are the primary benefits and limitations of using this approach? |
|
|
Three Options for Last Exercise |
|
|
Option One: Revenue Management Pricing |
|
|
Revenue Management Pricing |
|
|
Read: |
|
Mon April 23
Buc 103 |
Revenue Management Pricing cont’d |
|
1. Read the Forte Hotel Revenue Management Exercise (pp. 455 - 57) and explore the Yield Management Software: C:\Program Files\MktgEng\tutorials\revenue.pdf 2. The demand for a hotel room is a multiplicative function of price D = a* P-b a. Use the data in http://www.business.utah.edu/~mktbm/mkt6600/hotel.xls to estimate a and b for discount rooms. b. Estimate a and b for discount rooms judgmentally from the following information: It is expected that you can fill 90 rooms at a price of $55 and the price elasticity is –2.5. You should get similar information using the two methods. c. Estimate a and b for standard rooms judgmentally from the following information: It is expected that you can fill 50 rooms at a price of $130 and the price elasticity is –1.5. 3. Assume the marginal daily cost of a standard room is $13 and the marginal cost of a discount room is zero – i.e., the cost of maintaining a discount room is fixed. First, assume the hotel has 100 standard rooms. What is the optimum price for these rooms? 4. Assume that a single hotel with a capacity of 100 rooms can have a combination of standard and discount rooms, how many of each should it have (assuming the same fixed cost) and what price should be charged for each? |
|
Wed April 25
|
Revenue Management Pricing Application |
Exercise:
|
Forte Hotel Yield Management (pp. 455 - 57) 1. Answer the questions following the case. 2. For questions 5 and 6, reset the upper limit to price of $200 and the max difference in price to 12% so you are back to room prices of $200, $140, and $125. 3. For question 5, assume this is occurring 20 days before the target date and in question 6 assume this is occurring 15 days before target date. |
|
Option Two: Pricing for Value |
|
|
Pricing for Value |
|
|
Read: |
|
|
Mon April 23 Buc 103 |
Value in Use Pricing cont’d |
|
2. Read the Account Pricing for the ABCOR2000 Exercise (pp. 449 - 450) and explore the pricing for value software: C:\Program Files\MktgEng\tutorials\value.pdf software |
|
|
Wed April 25 |
Value in Use Pricing Application |
Exercise:
|
|
|
|
|
|
|
Option Three:
Promotional Decisions |
|
Mon April 18 |
Promotion Decisions |
|
Chapter 10 (pp. 435 – 48) |
|
|
Promotional Decisions |
|
|
Explore the Promotional Spending Tutorial (pp. 458 - 61) C:\Program Files\MktgEng\tutorials\massmart.pdf |
|
|
Promotion Decisions – Application |
|
|
MassMart Inc. (pp. 458 – 61) 1. What factors seem to influence brand choice and quantity purchased? 2. What is the optimal promotional strategy? For this strategy, what will be the composition of sales? What proportion comes from brand switchers and what proportion comes from purchase acceleration? What is the source of the additional profits? Both these are relative to the base case of no promotional activity. 3. Does it make sense to promote two brands simultaneously? 4. How do the trade deals (profit margins) affect retailers’ promotion decisions (which brand and how much?)? Should trade deals always be passed on? 5. Should this approach be adopted for all Mass Mart stores? |
|
|
Synopsis and Review |