University of Washington MBA Program

1
University of Washington MBA Program

• Managing Customer Relationships
• through Direct Marketing
• Financials and Budgeting
• Instructor Elizabeth Stearns

2
Simplified Example 20 Order
Average order, gross 22.00 Less returns
(10) 2.00 Average order,
net 20.00 Merchandise cost
(35) 7.00 Order processing 1.50 Return
processing loss 1.00 Overhead
(12.5) 2.50 12.00 Contribution to promotion
cost profit 8.00 40
Contribution 40 (8.00)
Net Sales 100 (20.00)
Promotion Cost 25 (5.00)
Profit 15 (3.00)
3
Break-Even and Profitability 20 Order
Contribution 40 8.00
Promotion cost target 25 5.00
Target profit 15 3.00
Magazine Ad Solo Mailing Catalog
Promotion cost per 1000 16 400 500
Orders per 1000 required to break-even 2.0 .2 50 5.0 62.5 6.25
Net sales per 1000 40 1000 1250
Orders per 1000 required to make target profit and advertising cost 3.2 .32 80 8 100 10
Net sales per 1000 64 1600 2000
Break-Even promotion cost per 1000 / 8 25
selling cost target promotion cost per 1000 /
5 Net sales per 1000 orders 20
4
R-F-M

• Recency
• When was the last time they purchased?
• Frequency
• How often do they purchase?
• Monetary Value
• How much money do they spend?

5
RFM Definition
RFM is a behavioral segmentation technique

• Typically used to select likely profitable
  customers to receive direct marketing treatment
• It postulates that the most likely prospects are
  recent purchasers who have historically
  demonstrated more frequent than average purchase
  behavior in larger than average dollar amounts
• It is based on a correlation between RFM and
  response
• Historically it has proven to be an effective
  segmentation technique for many situations
• The variables RFM are frequently influential in
  many advanced statistical modeling techniques

6
RFM Definition (contd)

• The variables must be interpreted within the
  context of product purchase dynamics
• Durable
• Consumables
• Periodic
• Can be used to manage marketing investment by
  selecting target customers
• Can also be used to improve marketing performance
  by managing message/offer components

7
RFM Elements
Segmentation Concept Behavioral Question Data Element/ Measurement
Recency When did they last buy? Date of last purchase Process Sort by date Create groups by date range
Frequency How often do they buy? of purchases over time Options Within recency group, months on file, times mailed
Monetary Value How much do they spend? value of purchases Options Within recency group, months on file, times mailed
8
RFM / Customer Treatment

• RFM segmentation can be used to manage customer
  treatment across functions
• Functional Dimension Treatment Implication
• Customer Service Service Level
• Billing Adjustment Practices
• Collections Minor delinquency action
• Credit Over limit action
• Marketing Incentives/Premiums

9
Gains Chart for Responsiveness
Decile Qty. Mailed (000) Number Resp. () Percent Resp. () Resp. Gain Index Revenue Generated (000) Mailing Cost (000) Total Profit (000)
1 600 20,250 3.37 225 506 240 266
2 1,200 34,200 2.85 190 855 480 375
3 1,800 48,330 2.68 179 1,208 720 488
4 2,400 60,840 2.53 169 1,521 960 561
5 3,000 67,500 2.25 150 1,687 1,200 487
6 3,600 76,680 2.13 142 1,917 1,440 477
7 4,200 81,900 1.95 130 2,047 1,680 367
8 4,800 84,960 1.77 118 2,124 1,920 204
9 5,400 87,480 1.62 108 2,187 2,160 27
10 6,000 90,000 1.50 100 2,250 2,400 -150
Mailing Cost 400/thousand Revenue
25/response
Index of Relative Responsiveness Universe
Indexed to 100. Source Direct Marketing,
November 1988
10
Sensitivity Analysis

• Three variables are often used to develop a
  picture that gives a range of possibilities to
  achieve financial objectives
• Response Rate (and the cost to achieve this)
• Average Order Size (and requisite merchandise)
• Circulation/Audience Size (and economies for
  volume)
• These are reviewed to give a reality check on
  Best/Worst/Most Likely scenaria, and provide a
  good basis for a monthly Cash Flow analysis.

11
Sensitivity Analysis Example
12
Cash Flow Analysis

• When looking at a start up, you need to look at
  monthly cash flow. You only have one year to
  succeed!!
• In Direct Marketing this is effected by, among
  other things (please refer to text)
• Inventory assumptions
• Fulfillment forecasts
• Additional bribes
• Response curves
• Cost of Capital

13
Lifetime Value of a Customer

• The net profit that you will receive from
  transactions with a given customer during the
  time that this customer continues to buy from
  you.
• Can help you make marketing strategy decisions
• If an investment increases lifetime value, do it!
• If an investment decreases lifetime value, dont!
• Often the benefits of a marketing investment do
  not come in the first year. This does not make
  it a bad investment!
• Three ways to improve lifetime value
• Increased retention
• Increased spending rate
• Increased referrals

14
Customer Lifetime Value
Year 1 Year 2 Year 3 Year 4 Year 5
Revenue
A. Customers 1,000 400 180 90 50
B. Retention Rate 40 45 50 55 60
C. Avg. yearly sales 150 150 150 150 150
D. Total revenue 150,000 60,000 27,000 13,500 7,500
Costs
E. Cost 50 50 50 50 50
F. Total costs 75,000 30,000 13,500 6,750 3,750
Profits
G. Gross profit 75,000 30,000 13,500 6,750 3,750
H. Discount rate 1 1.2 1.44 1.73 2.07
I. NPV profit 75,000 25,000 9,375 3,902 1,812
J. Cumulative NPV profit 75,000 100,000 109,375 113,277 115,088
K. Lifetime value (NPV) per customer 75.00 100.00 109.38 113.28 115.09
15
Table 10-1. Mary Annes Closet
Revenue Year 1 Year 2 Year 3
R1. Customers 10,000 3,000 900 (3)
R2. Retention Rate 30.00 30.00 30.00
R3. Spending Rate 120 120 120
R4. Total Revenue 1,200,000 360,000 108,000 (4)
Variable Costs
C1. Percent 70.00 70.00 70.00
C2. Total Variable Costs 840,000 252,000 75,600 (5)
Profits
P1. Gross Profit 360,000 108,000 32,400 (6)
P2. Discount Rate 1.00 1.16 1.35
P3. NPV Profit 360,000 93,103 24,000 (7)
P4. Cumulative NPV Profit 360,000 453,103 (1) 477,103 (8)
L1. Customer Lifetime Value 36.00 45.31 (2) 47.71 (9)
1. 360,000 93,103 2. (1) / 10,000 3.
3,000 .3 4. 120 (3) 5. 0.6 (4) 6. (4)
(5) 7. (6) / 1.35 8. (1) (7) 9. (8) / 10,000
16
Table 10-2. Mary Annes Closet with the Birthday
Club
Revenue Year 1 Year 2 Year 3
R1. Referral Rate 8 8 8
R2. Referred Customers 0 (1) 800 (5) 464
R3. Retained Customers 10,000 5,000 2,900
R4. Total Customers 10,000 5,800 (6) 3,364
R5. Retention Rate 50 50 50
R6. Spending Rate 150 150 150
R7. Total Revenue 1,500,000 870,000 504,600
Variable Costs
C1. Direct Percent 70 70 70
C2. Direct Costs 1,050,000 609,000 353,220
C3. Birthday Club Mailing Gift 50,000 (2) 29,000 16,820
C4. Birthday Discounts _at_ 4 40,000 (3) 23,200 13,456
C5. Referral Gifts _at_ 5 0 (4) 4,000 2,320
C6. Total Costs 1,140,000 665,200 385,816
Profits
P1. Gross Profit 360,000 204,800 118,784
P2. Discount Rate 1.00 1.16 1.35
P3. NPV Profit 360,000 176,552 87,988
P4. Cumulative NPV Profit 360,000 536,552 624,540
L1. Customer Lifetime Value 36.00 53.66 62.45
Assumptions
1. Assume referrals buy in Year 2. 2. 5 per
customer mailing balloons. 3. 20 will
use discount, spending average of 50 times
20 discount divided by total customers
on database 4. No referrals in Year 1. 5.
Referral rate of 8 percent times Year 1
total customers. 6. Retained customers plus
referred customers.
17
Net Change in Customer Lifetime Value
Year 1 Year 2 Year 3
Before the Club 36.00 45.31 47.71
After the Club 36.00 53.66 62.45
Change 0.00 8.35 14.74
Times 50,000 Customers 0 417,500 737,000
18
Modeling Definition

• A model can be thought of as an equation that
  predicts an outcome
• A wide range of statistical techniques can be
  employed
• Can consider the value of all available data to
  predict an outcome
• It is a rigorously disciplined approach to
  applying information to the marketing process

19
Modeling / Scoring Process

• In building a model
• An appropriate dataset is defined or created
• The variables that explain the outcome are
  determined
• The explanatory power, weight of each variable is
  identified
• The way the variables work together is studied
  and understood
• The variables are then used to build the model
  equation which is then balanced, tested, and
  tuned to best fit and predict the desired outcome

20
Modeling / Scoring Process

• Scoring defined
• Scoring is the process of applying a model
  equation to each observation or customer record
• This is a mechanical process of looking at the
  model variable for each customer
• Applying the weights for each variable
• Calculating a numeric value for each customer
• Using the resulting value to rank each customer
• The ranked customer list is then divided into
  groups containing equal numbers of customers,
  often 10 groups, sometimes more
• The top declines contain higher percentages of
  the best prospects and therefore perform better
• The resulting increase in performance of these
  groups over the average is called lift or gain

21
Database Marketing

• Over the past few years, database marketing has
  been a very hot topic
• The cost of computer performance has rapidly
  decreased, and the functionality of software has
  increased
• Knowledge of how to apply statistics to direct
  marketing has grown
• Today, systems are being used to target
  prospects, profile customers, manage customer
  relationships, model behavior and measure risk
• Database marketing is a prerequisite for success
  in the 90s

22
Elements of a Marketing Database
Psychographic Overlays
Geodemographic Overlays
External Overlays (Adverbs)
Contact History
Response History
Marketing Mgmt Info (Adjectives)
Promotional History
Performance Metrics (Verbs)
Monetary Value
Recency
Frequency
Customer Definition (Nouns)
Customer/Prospect
23
Database Needs Analysis

• There are a variety of technical solutions that
  may be used to address database marketing
  requirements
• Considerations include
• Applications
• Data Availability internal and external
• On-line access
• Data retention
• Operations
• Decision Support
• Frequency of data changes or updates
• Existing hardware and software
• Organization and personnel
• Security

24
Marketing Management Information
Customer Data

• Customer
• Characteristics
• Demographics
• Psychographics
• Household Composition

Contact History
Customers/ Prospects Customer-Level View
Response History
Behavior Change Drivers
Promotion History
Key Behavior Events
Selection System
Reporting And Analysis System

• Performance Metrics
• Recency
• Frequency
• Monetary Value

Reports/ Evaluation
Program Streams
Pre-Purchase Programs
Retention Programs
25
Response Modeling Process
Prepare Research Files
Develop Response Model
Score Files
Rank Files by Descending Score
Divide Files into Deciles
Calculate Gains Tables
26
Prepare Research Files
Customer File
Take sample of customer file for research
Research Sample
Append enhancement data such as demographics and
response history
Recode variables and calculate new variables
(such as response rate)
Enhanced Research Sample
27
Prepare Research Files
Split sample into an analytic file and a
validation file
Enhanced Research Sample
Analytic File Used to build response model
Validation File Used to check accuracy of
response model
28
Develop Response Model
Build Regression Model
Regression Equation
Y Intercept B1Var1 B2Var2 BnVarn
Where Y Probability of Response Bi Linear
regression coefficient for variable I Vari
Variables describing attributes of individual
customers
29
Develop Response Model
Example
Intercept
Y 0.257 0.069 Resrate 0.072
Income range 0.009 Age 0.159
Male
Variable Description Resrate Response rate to
prior promotions Income Range Income (recoded
range is 1 to 4) Age Age of customer Male Indi
cates customer is male
30
Develop Response Model
Select best model based on Regression
statistics Analytic and validation gains
tables Intuitive appeal
31
Score Analytic File
Apply equation to every customer record on file
Example
Customer 1 Customer 1 Customer 2 Customer 2
Variable Coefficient Value Variable Value Partial Score Variable Value Partial Score
Intercept 0.257 0.257 0.257
Resrate 0.06915 0.25 0.017288 0.012 0.0008298
Income Range 0.07152 4 0.28608 2 0.14304
Age -0.00875 48 -0.42 25 -0.21875
Male -0.159 0 0 1 -0.159
Total Score 0.14037 0.0231198
32
Score Analytic and Validation Files
Regression Equation Y 0.257 0.06915 Resrate 0.
07152 Income Range -0.0875 Age -0.159 Male
Analytic File
Scored Customer File
Repeat scoring procedure for validation file
33
Rank Files by Descending Score
Analytic
Highest Score Most likely to respond
Customer 1
Customer 2
Lowest Score Least likely to respond
Repeat ranking for validation file
34
Divide Files into Deciles
Analytic
Highest Score Most likely to respond
Customer 1
Customer 2
Lowest Score Least likely to respond
Follow same procedure for validation file
35
Calculate Gains Tables
Analytic
Cumulative Cumulative Cumulative
Decile Mailed Resp Rate Mailed Resp Rate Cutoff Score
1 650 85 13 650 85 13 0.125
2 650 60 9 1300 145 11 0.098
3 650 45 7 1950 190 10 0.086
4 650 29 4 2600 219 8 0.073
5 650 26 4 3250 245 8 0.069
6 650 15 2 3900 260 7 0.054
7 650 11 2 4550 271 6 0.045
8 650 5 1 5200 276 5 0.035
9 650 4 1 5850 280 5 0.019
10 650 1 0 6500 281 4 ----
36
Calculate Gains Tables
Validation
Cumulative Cumulative Cumulative
Decile Mailed Resp Rate Mailed Resp Rate Cutoff Score
1 650 71 11 650 71 11 0.125
2 650 53 8 1300 124 10 0.098
3 650 40 6 1950 164 8 0.086
4 650 32 5 2600 196 8 0.073
5 650 30 5 3250 226 7 0.069
6 650 27 4 3900 253 6 0.054
7 650 12 2 4550 265 6 0.045
8 650 6 1 5200 271 5 0.035
9 650 4 1 5850 275 5 0.019
10 650 5 1 6500 280 4 ----
37
Produce Gains Table for Entire Customer File
Score customer file using regression
equation Rank file by score Divide file into
deciles Calculate gains table
38
Customer File Gains Table
Incorporating Financial Data
Cumulative Cumulative Cumulative Cumulative Net Total Profit
Decile Mailed Resp Rate Mailed Resp Rate Cutoff Score Cumulative Net Total Profit
1 6500 854 13 6500 854 13 0.125 11,517.00
2 6500 602 9 13000 1456 11 0.098 17,238.00
3 6500 455 7 19500 1911 10 0.086 19,578.00
4 6500 299 5 26000 2210 9 0.073 18,330.00
5 6500 264 4 32500 2474 8 0.069 16,277.00
6 6500 158 2 39000 2632 7 0.054 11,786.00
7 6500 112 2 45500 2744 6 0.045 6,237.00
8 6500 53 1 52000 2797 5 0.035 (669.00)
9 6500 41 1 58500 2838 5 0.019 (7,851.00)
10 6500 6 0 65000 2844 4 ---- (15,838.00)
Net total profit (Number of responders
average profit per response) -
(Number of pieces mailed total cost per mailed
piece)
Total cost per mailed piece 1.25 Profit per
response 23.00
39
(No Transcript)
40
(No Transcript)