Segmentation Applications in Banking - Data Mining Techniques in CRM: Inside Customer Segmentation - page 258

A ''fashion'' card with a relevant reward scheme was the bank's proposition
for the ''Fashion ladies'' of segment 6. The rewards tied to card usage included
rebates and zero-rate installments for purchases at retail affiliates. This card was
advertised in women's fashion magazines. Stands with card brochures were also
positioned in affiliated fashion stores. A bargain fashion catalogue was also mailed
to members of this segment, promoting card purchases at partnering stores.
''Fun lovers'' seemed perfect candidates for a ''student'' card. The rewards of
this card were tied to card usage at restaurants, bars, clubs, cinemas, bookstores,
and so on. The card was advertised in magazines, on TV shows, and on web
sites that were popular among young people, the advertised message presenting a
modern and cool brand image.
A final type of card, designed and developed to target the ''Occasional users'' of
segment 8, offered reductions and zero-rate installments when used at partnering
retailers of electrical appliances and electronics. This card could also be issued by
these authorized retailers on behalf of the bank.
BEHAVIORAL SEGMENTATION REVISITED: SEGMENTATION ACCORDING
TO ALL ASPECTS OF CARD USAGE
After segmenting customers according to their purchasing habits and the products/
services they spend their money on, the bank also realized the need for a second
segmentation which would focus on general usage aspects such as frequency and
intensity of spending, type of spending (purchase or cash advance), and payment
behavior.
Although the bank could have used all the available purchase and usage
data in an integrated segmentation model aiming to cover all behavioral aspects,
it turned out that implementing two separate and more targeted segmentations
provided a clearer separation. In the end, the two distinct segmentation schemes
were combined in order to provide a unified profile of each customer in terms of
both general usage and purchasing habits.
The general usage segmentation encompassed all the usage fields listed in
Table 6.2. Once again the input dataset covered a time period of six months. The
segmentation population included all active credit card holders, regardless of their
monthly purchase amounts. The basic data dimensions were uncovered using a
principal components model before the application of cluster analysis. To make
a long story short, the main steps of the procedure were analogous to the ones
presented in the previous project. So we will skip these steps and proceed directly
to the derived results and the profiling of the revealed clusters.
The selected segmentation scheme was composed of six customer groups.
Table 6.11 presents the means of some key usage indicators by cluster.