Eric T' Bradlow

1
Developing a Market-Driven Strategy in the Air
Filtration Industry

• Eric T. Bradlow

Associate Professor of Marketing and
Statistics The Wharton School
2
A Smarter Consumer?

• The increased utilization of multiple retailers,
  the internet, national distributors has led to
• Increased knowledge availability and usage,
• Greater strain on profit margins (consumers are
  shopping around),
• A need to consider new ways to attract and retain
  customers.

Companies need to carefully evaluate their value
proposition to consumers
3

• We will focus in this session on three key areas
• Methods and Value of Customer Acquisition
• Do I really want your business?
• And how much will it cost me to get it?
• Customer Loyalty and Retention Programs
• What will it cost me to keep you?
• Dollar Volume Extraction
• Can I get a greater dollar volume of your wallet
  share?

4
A Change in Philosophy

• More and more companies are relying on classic
  marketing strategies/calculations to achieve
  these goals.
• Segmentation
• Loyalty Retention Programs
• Lifetime Value of the Customer
• Cross-Selling via Data Mining

5
Issue 1 How do we identify which customers to
try and acquire?
Which ones are the keepers?
6
Basic Premises

• It may be costly to acquire new customers.
• There are many customers whose costs outstrip
  their revenue.
• Market Segmentation is a classic approach to
  bundling customers.

7
The Essence of Marketing is Discrimination

• Propositions (Are these true for Camfil Farr?)
• People like to have things that exactly meet
  their needs, as opposed to things that function
  OK, but do not precisely meet their needs
• People like to be talked to in ways that
  reflect full understanding of their needs, rather
  than in more general terms
• People will often pay more for things that
  exactly meet their needs, and/or as a result of
  being talked to in specific ways (ABILITY TO
  EXTRACT SURPLUS)

8
Everyone is Different!!
However, it is infeasible to target each
individual customer.
9
Implementing Segmentation Steps

• Market Segmentation
• Identify bases for segmenting the market
• How do you decide who is in what segment?
• Develop profiles or segments
• Using Statistical Clustering Routines.
• Relate to descriptors for reachability (who are
  these people?)

IMPLEMENT
10
Major Segmentation Variables

• Demographics (age of accounts, urban/rural)
• Usage (high volume, types of accounts)
• Purchasing Approaches (internet, agent,
  longstanding clients)
• Needs (What do they want?)

Which is most appropriate for Camfil Farr? Is
the Clean Air, Comfort Air, Safety the right
segmentation?
11

• Target Marketing
• Develop measures of segment attractiveness
• dollar volume usage, profit margin, etc.
• competitive landscape for these customers
• LOYALTY (will discuss later)
• Typically an attractiveness matrix will help to
  evaluate these alternatives.
• The new rage is CUSTOMER METRICS.
• 5. Select the Target Segments

12
Attractiveness Matrix
Segment Loyalty
High
Low
Margin
High
Low
High
Low
?? value
profits ?? Costly to get
Can Obtain! Defend?
Large Group, can you retain?
Big
Size
Dont play
Niche Player
Worth it?
Niche Player
Small
Where does Camfil Farr live? Where should you
live?
13
Customer Metrics

• Loyalty Score
• Years as customer
• Number of accounts
• Number of consecutive transactions
• Volumetric Scoring Model (combination of dollar
  volume and loyalty).
• Lifetime Customer Value

14
Lets Observe the Event Histories
X
n4
A
X
X
X
t0.9
B
X
X
X
X
n4
t0.3
C
X
X
X
X
X
X
n6
t0.5
D
X
X
n2
t0.5
1
0
12/31/01
Four Customer Event Histories
1/1/01
15
Intuition Check

• Which of the four customers would you like to
  have going forward?
• Which of the four customers is most likely to
  purchase again in 2002? Which customer is least
  likely?
• What are the key drivers of customer value here?

16
Issue 2 Now that we have decided on the
attractive segments, how do we keep the people as
our customers.
Retain and Defend
17
Loyalty Retention Programs

• The success of virtually every product is
  determined by REPEAT and not initial client
  behavior.
• Loyalty retention programs
• Bonus points.
• Free Accounts.
• Mileage.
• Sweepstakes.
• Rate discounts (????), introductory.
• Partnering Agreements.

Would these things work for Camfil Farr?
18
Relevant Calculation

• For EACH target segment you must compute
• Cost of loyalty program (C)
• Probability of depth of repeat usage with and
  without the loyalty program P(t), Q(t)
• Dollar volume usage with and without loyalty
  programs V(t), W(t)
• Estimated segment size N

19
Ways to Obtain this Data

• Historical Data
• Direct Experimentation
• Survey your customers
• Managerial Expertise (Guesstimation)

20
Loyalty Program Value

• LPV
• NLPV
• Choose one with greater value (maybe).
• For example, introductory price programs are
  something that should be evaluated carefully.

21
Assessment of Risk

• Present value of alternatives is one
  consideration.
• Risk of alternatives is another.
• When do I start seeing the profits?
• What are the critical assumptions (scenario
  analysis)?
• What will my competitors do
• Prisoners Dilemma

22
Issue 3 How do I get a larger portion of my
clients dollar-wallet share?

• Cross-Selling

23
Facts about Cross-Selling

• Most companies by their organizational structure
  are woefully prepared for cross-selling.
• E.g., organized by product line.
• No overall management team.
• Many companies do not share database information
  across product lines and/or have incompatible
  software.

24
The Value of Cross-Selling

• Cross-Selling is a very inexpensive way to
  increase revenue
• You already own the customer
• Increases customer loyalty
• Lower distribution costs
• Lower acquisition costs
• Thwarts inroads by competitors

25
Data Mining as a Tool for Cross-Selling

• Example
• Customer A 30/30
• Customer B 30/30, Citysorb
• Customer C XH Absolute Filters
• Customer D 30/30, Durafil

26
Uses of Data Mining

• Who do I target market for 30/30s?
• Which bundles of products should I offer to
  customers?
• Which products are complimentary/substitutes?

27
Data Mining Answers

• Data Mining techniques allow one to efficiently
  search (and fit models to) large data bases using
  sets of rules and table look-ups
• Which product is most often bought in conjunction
  with the 30/30?
• Which bundles of air filters are typically bought
  together?
• Which products purchase is the antecedent to a
  given products purchase?

28
Summary and Final Thoughts

• Identify the profitable customer segments.
• Market segmentation.
• Evaluate the effectiveness of loyalty programs.
• Compute NPV and assess risk
• Scenario Analysis for Risk Assessment.
• Obtain larger dollar share using cross-selling
  techniques (DATA MINING)
• Efficient, increased loyalty, and coherent
  marketing strategy.

29
Profits are primarily made from extracting a
surplus from added-value features. However,
people must be willing to pay, and willing to
stay.