Download From Data Modeling to Business-to-Business Application Development: Targa Services/Fiat Auto Case

From Data Modelling to
Business-to-Business
Application Development :
Targa Services / Fiat Auto Case
G. Lux
Targa Services
A. Dorigo
Nunatac
A. Saccardi
Nunatac
Who’s Who ?
Group Company
created to develop and manage
services for Fiat customers
through the dealers’ network.
Indipendent consulting firm
focused on delivering tailor
made business solutions for
decision support and CRM.
Targa Services
DEALERS
•Customer DB
•Marketing Tools
•Campaign Planning
•Breakdown Service
•Call Centre
•Direct Mailing
•Sales (new/used)
•Service
CUSTOMERS
A bit of history…
mid-1996
Targa Services starts the development of an
entity relational Data Warehouse for Fiat,
Lancia and Alfa Romeo customers.
end-1997
TS releases MIDA (Marketing Integrated
Data Base) and starts the integration of
external data about prospect customers.
mid-1999
TS chooses Nunatac to design and implement
a Data Mart for analytical purposes. SAS
supplies the software technology.
Level 1: MIDA Data Warehouse
SALES
CALL
CENTRE
LOANS
•ORACLE DB
EXTERNAL
LISTS
INSURANCE
•OPERATIONAL DATA
SOURCES
SECOND HAND
BUY BACK
SYSTEM
•NORMALIZED
WARRANTY
BREAKDOWN
SERVICE
•LOW GRANULARITY
SERVICING
RENT A
CAR
Level 2: Data Mart for Customer Analysis
WHAT ?
WHY ?
SAS Data Base comprising
lower detail data tables (Star
Schema) and summarized data
tables (Customer Table) for
customer profiling
To enable quantitative analysis
aimed at transforming the set of
customers and prospects in a
dynamic target for relational
marketing campaigns
System Architecture
DMA
PL/SQL
SAS
Detailed
Summarized
MIDA
DMA
1 record for
each customer
car
private
loans
00101001100 00 1 00 1 0101 1 0
warranty
C.C…Service…Targa Ass...
REPOSITORY
STAR SCHEMA
CUSTOMER TABLE
The information stored in the DMA
MIDA is the only data source of the DMA.
MIDA
DMA
Fiat Customers
Regular Update
Private Data
Car Ownership
Optional
Products/Services
Contacts
Competitors’
Semi Annual Upload
Private Data
Car Ownership
?
Contacts
Technicalities…
Number of Tables:
14 + 14
Overall Dimension:
30 GB
Integral Scan:
yes
Update Frequency:
2 weeks
Elapsed Time:
22h
Server:
DecAlpha 8400
CPU:
6 x 524 Mhz
A few numbers…
Cars Customers
(Individuals + Businesses)
Services Customers
(Individuals + Businesses)
Contacts
(Outbound+Inbound)
Total Cars
21.900.000
3.100.000
6.760.000
29.600.000
…and now that you’ve got this DMA?
•
Customer ID Card
•
Loyalty and
Transition Matrix
•
Car Ownership GeoDistribution
•
Campaign Reporting
•
Return on Marketing
and Sales
Customer ID Card
For EVERY
CUSTOMER
There is…
Structural Data
+
+
=
A quantitative description of
the relevant variables for
marketing purposes
Consumer, Age,
City…
Original Behavioral Data
Tot Cars, Avg. Cost,
Purchase Freq…
Derived Behavioral Data
Custom, Brand Loyalty,
Expt Next Purchase…
Customer ID Card
Customer id:
0009348074
Age:
47
Sex:
Male
Cars Total:
4
Fiat Group:
2
Last Brand:
FIAT
Last Model:
Panda
Loyalty Index:
90
Purchase Forecast:
Oct 2000
Loyalty and Transition Matrix
Taken the car sales of a semester, the matrix compares the
brand of the last purchase with the previously owned one.
For example…
1,1 % of new A lfa
owners com e from
BM W
1,5 % of old A lfa
owners buy an A udi
in the sem ester
C_BM W
3,9
20,8
4,6
R_BM W
217 1,5
1162 24,5
257 6,4
N_BM W
41,3
1,2
1,1
C_AUDI
29,3
2,5
2,8
R_AUDI
C_ALFA_R
4230
120
115
N_AUDI
R_ALFA_R
MAR_OLD
ALFA ROMEO
AUDI
BMW
N_ALFA_R
120 custom ers
who owned an
A udi bought and
A lfa Rom eo in
the period
MARCA NEW
276 1,9
362 7,6
2004 49,6
5,6
7,4
41,0
2004 BM W
custom ers are
loyal to their
brand in the
period considered
Car Ownership Geo-Distribution
Area Code:
00060
City/Town:
Sacrofano
Tot. Residents:
7.635
Tot. Families:
2.098
Tot. Car:
1.899
Tot. Gr. Fiat:
876
Tot. Elite:
109
Average Age:
5,7 anni
Etc…
For every single geo unit of the Italian territory, the DMA
contains informations about the cars owned by its residents.
(max granularity ISTAT zone)
City/Town Id:
10-037-006
Zone:
0000766
Individulas:
346
Families:
145
Car Owners:
129
Car (new):
115
Car (used):
26
Car (mkt.seg.A):
12
Car (mkt.seg.B):
56
Etc.. Etc..
One-to-One Campaigns
Targa Services supports
Fiat, Alfa Romeo and
Lancia Brands and
Dealers with tailored
CRM solutions
•
It helps the dealer
choose the
appropriate target
•
It provides the list of
customers’ addresses
3. It supplies the customized mailing package
for the specific campaign
Campaign Reporting (Web based)
CAMPAIGN
PLANNING
TARGET
SELECTION
MEDIA
PLANNING
CAMPAIGN
ROLL-OUT
CAMPAIGN
REPORTING
TEST DRIVE,
VISIT C/O DEALER
…
SALES
START
Return On Marketing and Sales
DEALERS’
CUSTOMER DATA
BASE
Improving
customer and
prospect
management
ROMS
Any Question?
From Data Modelling to Business-to-Business Application
Development: the Targa Services / FIAT Auto Case.
Giovanni Lux – Targa Services (FIAT Auto)
Andrea Dorigo, Alberto Saccardi – NUNATAC
Abstract
Database Driven Direct Marketing is not just a matter of implementing an efficient Customer
Database, but involves the process of developing customized analysis tools and sharing marketing
intelligence in a Business to Business perspective.
To meet these requirements you need:
! A dedicated Customer Database (CD), structured for data analysis, containing information on
identity and buying behaviour of all your customers;
! Skills and tools for producing relevant market and customer intelligence from the CD;
! An application for managing outbound and inbound information flows while delivering up to date
campaign reporting;
! A system which integrates Central Sales Dept. with Dealers for sharing marketing strategies.
Targa Sevices is a FIAT Auto Group Company whose mission is to introduce innovative customer
oriented marketing solutions.
In this paper we describe the joint Targa/Nunatac effort to achieve the above mentioned objectives
with the purpose of showing how this approach gives sounder knowledge of your customers and leads
to the development of better performing Business to Business marketing tools.
Introduction.
(ITA) with offices in the main European
capitals;
Local Car Dealers spread all over Europe.
Nunatac is a Milan (ITA) based consulting
firm which operates in the Italian market
through a team of experienced professionals
with statistical, computing and marketing
skills. Focused on delivering tailor made
business solutions for decision support,
Nunatac is a SAS Institute European Quality
Partner and has implemented various projects
in the field of Data Warehousing, Data Mining
and Database Marketing exploiting SAS
System functionalities.
!
Targa Services is a company of the FIAT Auto
Group created in 1996 with the aim of
providing FIAT customers everything they
need apart from the car itself: loan, insurance,
warranty extension, traffic and tourist
information,
additional
services
etc.
Nevertheless, the mission of Targa Services is
also that of improving customer relationship
management by providing relevant information
on market trends, customers’ behaviour and
prospects’ needs. Strictly speaking, the clients
of Targa Services for this kind of added value
information fall into two main categories:
! Marketing users of the Central Sales and
Marketing Department based in Turin
Fiat Auto commitment to build a Customer
Database began in 1995 and the project has
been carried out through the creation, in 1996,
of Targa Services which released the very first
version of MIDA (Marketing Integrated DAta
base) in late 1997. The initial goal of MIDA
was to collect from the operational systems,
and organize in an entity relational database, as
many data as possible about the customers of
the Fiat Group three brands, namely FIAT,
Alfa Romeo and Lancia. That means not only
data about car ownership, but also about all the
different services provided by Targa itself.
As a second step in the implementation of the
database, Targa Services started the integration
of additional data from external lists to be
It’s in the scope of this paper to present the
solution developed by Targa Services and
Nunatac to fulfil the information requirements
of both these two categories of users giving
them a description, as accurate as possible, of
their customers while taking into account their
different business needs.
1. The Existing System.
1
added to those already stored. The final result
of this integration was an Oracle database
stored on a DecAlpha/UNIX server and
accessible to users from the Central Sales
Department via SQL server.
Sales
Network
Regular and Extended
Warranty
Orders and
Deliveries
Used Cars Warranty
Check Up
Program
Road
Assistance
Loans Data
MIDA
Sales to
Employees
The DMA was projected to have a clear logical
reference, the customer, and therefore to be
highly de-normalized.
The business requirement the DMA was
expected to meet, was basically the need to
have a comprehensive view of Fiat customers.
The idea was not only to know what customers
do in terms of car purchases (and use of car
related services) but also to understand why
they behave that way.
3. A Data Mart for analytical purposes.
Marketing Integrated Data Base
Second Hand
& Buy Back
System
Targa Service
Systems
Call Centre
Market
Sensitive
Rent a car
External Data
Insurance
2. The Business Perspective.
Despite these numbers, MIDA is an entity
relational, normalized database which can store
loads of data very efficiently but remains quite
unsuitable for analytical purposes.
For this reason, in early 1999 Targa Services
started a project with Nunatac to built a Data
Mart dedicated to customer analysis, called
DMA.
PL/SQL
Flat
Files
Star
Schema
Flat
Files
SAS System
Customer Table
Star
Schema
Customer Table
BUSINESSES
Geo-marketing, customer identification and on
demand mailings were also included in a later
release of MIDA.
***
With its 39 million customer records (27 for
the Italian market and 12 for the other 11
countries featured in the database) and 60
million for products/services, MIDA is the
largest source of information on the car market
in Italy and one of the largest in Europe. The
growth rate averages a 10% a year and the
approximate size in bytes has fast reached and
exceeded the terabyte.
MIDA
INDIVIDUALS
Further developments of the MIDA project led
to the opening of MIDA access to the network
of Fiat, Alfa Romeo and Lancia dealers all
over Europe in a business to business
perspective. The main functionalities available
to the dealers, via the intranet, were at this
stage:
! target selection (and download), according
to the rules established by Fiat Auto
regarding customers ownership;
! access to targets selected for “central”
marketing initiatives for recall purposes;
! access to the loyalty plan, managed by
MIDA on behalf of the dealers.
In compliance with the policy chosen by Targa
Services, the only source of data for the DMA
is the MIDA database and the granularity of
the first is the lowest available in the latter.
Due to the different type of target, in terms of
marketing strategies, it has been decided to
split the whole structure of the DMA between
tables concerning private car owners
(Individuals) and corporate car owners
(Businesses).
Furthermore, tables concerning different
categories of facts have been kept physically
separated so that, both at the Star Schema level
and at the Customer Table level we can
identify eight main business entities:
1.
2.
3.
4.
5.
6.
7.
8.
Cars
Private data
Contacts
Loans
Extended Warranties
Maintenance
Call Centre Cases
Road Assistance.
2
ROAD ASSISTANCE
MAINTENANCE
CALL CENTRE CASES
EXTENDED WARRANTIES
LOANS
CONTACTS
CARS
PRIVATE DATA
STAR SCHEMA
CUSTOMER TABLE
As in the figure above, the structure of the
DMA comprises two different levels:
☺ The STAR SCHEMA level in which is
stored the most accurate detail of data. At
this level, and for each of the two lines of
customers (Individuals and Businesses),
we find eight different tables, one for each
business entity.
☺ The CUSTOMER TABLE level in which
we find one single record for customer in
each of the eight tables.
The choice of keeping tables of different
business entities separated also at the second
level was meant to optimise the access to these
quite ‘large’ tables (having sometimes more
than 70 variables) even if, theoretically, you
could merge (by consumer) the eight tables
making one big Customer Table which
wouldn’t be very easy to handle.
These optimisation concerns became relevant
due to the size of the DMA itself. Up to now,
there are stored data ‘just’ on Italian
customers, nevertheless some of its tables have
several million rows. The DMA is refreshed
every other weekend and the method selected
for the refreshment has been the integral scan
of the whole tables of MIDA so that every two
weeks the DMA is completely re-generated.
Stars.
Car
fin
businesses
2.735.148
817.377
53.606
exw
435.742
49.669
win
2.901.068
664.239
11.160
762
cc
Ctab.
individuals
24.530.982
auto
20.130.820
741.084
anag
22.756.878
1.312.108
ctfin
781.371
35.813
5.948.984
289.026
cnt
ctexw
418.920
35.253
ctwin
1.571.900
117.848
9.546
718
ctcc
The Customer Table level of the DMA
contains mainly derived data, but, for each
customer, it was chosen to keep the detail of all
the events concerning the last two cars owned
by that customer. Therefore, for instance, in
the car customer table, one can find all the
various counters for the total number of
purchased cars and so on, but also the detail –
brand, model and version - of the last two cars
purchased (the very last one being the one
currently owned by the customer).
The reason for this redundancy is that, in a
business perspective, the events related to the
cars prior to the last two may not be relevant at
all. There is a huge difference between
Customer A who called the Customer Care
Service ten times for a technical failure
happened ‘three cars ago’ and Customer B who
called ‘just’ three times but for the car he
presently owns. The total counter, in this case,
might not give the right information if it
assigns an higher criticality degree to the
former instead of the latter.
***
At the present stage of the project, the
information stored in the DMA is available
only to the first of the two categories of users
listed above (Marketing users of the Central
Sales and Marketing Department). For these
users, Nunatac has developed a number of
index values, computed for each customer and
inserted in the car customer table, aimed at
describing the buying potential of the
customer, its loyalty to a specific brand (not
only Fiat Group brands are featured in the
DMA), the likelihood of a new purchase in the
months to come and so on.
A loyalty matrix was designed and
implemented for monitoring how the customers
who buy a car in each semester switch from
one brand to another or remain loyal to the
brand they used to own.
4. The Sharing of Information.
Thanks to the common belief of Targa Services
marketing managers that strategic information
available centrally should be shared with the
Dealers Network for a sounder and more
effective implementation of marketing actions,
a set of tools is being devised to allow each of
the 800 Italian Dealers to access on line the
portion of the DMA related to its own
customers. The tool is meant to be scalable,
giving the dealer the possibility of integrating
the DMA view with local data unknown
centrally (an example of this type of
3
information is the maintenance done after the
warranty period).
While making the DMA become a reality,
Targa Services got concerned about another
aspect of the overall effectiveness of its
marketing policies: the redemption of
marketing campaigns.
For this specific purpose, in the second half of
1999, Nunatac, supported by SAS Institute
consultants, started the development of a new
Web based application for Campaign
Reporting.
5. The Web based Campaign Reporting.
From this project Targa Services was
expecting a tool, user friendly and Web
integrated, through which both the central
users and the local dealers could monitor the
performance of marketing for a certain
campaign by measuring the redemption of the
specific type of action selected.
MIDA/DMC
sales
SAS Intrnet
campaigns
contacts
MDDB
server
MDDB
report
viewer
The application, released by the end of 1999,
works with the SAS MDDB server which
generates a data hypercube putting together
information about car sales and outbound
campaign contacts and computing the
redemption of the campaign. This data is
stored on a dedicated Oracle data mart, part of
MIDA itself, which is called DMC (Campaign
Data Mart) and is retrieved with SAS Access
to Oracle. Users can access the Reporting on
line via intranet and navigate the hypercube
thanks to SAS MDDB Report viewer.
Due to security concerns, the solution adopted
for the Dealers was to create as many
hypercubes as the number of the dealers so that
every one of them could drill down its own
particular cube of data without interfering with
other dealers and affecting the performance of
the system.
In the transition to SAS V8, this solution isn’t
viable anymore and Nunatac is devising, in
collaboration with SAS Institute Technical
Support, a brand new way for handling security
with an unique hypercube accessed
simultaneously by different users with different
rights.
6. Conclusions.
The Data Mart for Customer Analysis (DMA)
and the Web based Campaign Reporting are
two milestones in the Fiat Auto path to an
improved Customer Relationship Management.
The deepest understanding possible of
customer needs and expectations is in the
philosophy of Targa Services and these two
projects have been conceived to provide a set
of state of the art marketing tools for achieving
this objective.
Web based technologies are significantly
influencing the development of these tools,
making possible the sharing of strategic
information between the Centre and the Local
Dealers in what is becoming more and more a
e-b2b perspective.
This will lead to a coordinated approach to the
business in which shared marketing strategies
will be implemented by the Sales Force in a
team oriented, proactive way.
Success is determined also by the right
technology.
In this paper we discussed a case in which SAS
products, from SAS Base and Stat themselves
to SAS MDDB Server and Report Viewer,
played a key role in shaping the whole project.
Nevertheless, translating data into relevant
information is a complex matter which requires
an expertise that goes beyond the mere
technology concerns.
This expertise is Nunatac’s mission.
Targa Services
Corso Settembrini, 215
10135 TORINO - ITALY
tel.
+39.011.6841778
fax.
+39.011.6841091
[email protected]
Nunatac
Via Crocefisso, 5
20122 MILANO – ITALY
tel.
+39.02.86996848
fax.
+39.02.89012074
http://www.nunatac.it
[email protected]
[email protected]
4
5