graduation thesis presentation, march 2012
DESCRIPTION
TRANSCRIPT
ALMA MATER STUDIORUM – UNIVERSITY OF BOLOGNA
Retention, Revenues and Multichannel Usage: an Empirical Analysis
Buzko Sergii 09000039719
Supervisor: Dr. Valentini Sara
Third session A.Y. 2010/2011
Research Methodology
2
The main questions 1) whether consumers who shop across all distribution channels
show higher spending behavior and/or stronger loyalty?
2) if so, which specific channel combinations (e.g. Internet-Store, Catalog-Store, etc.) are more profitable?
3) and how managers can exploit this opportunity to build a sustainable competitive advantage?
Models elaboration
3
hh
hhhhhhthh
DAgCISISCSCIOSOIOCv
ξββ
βββββββα
+++
++++++++=
h98
7654321
cs_hat
Re
o Revh = Annual revenue for customer h oChannel combinations a customer have used:
- OC = 1 if only catalog, - OI = 1 if only Internet, - OS = 1 if only stores, - CI = 1 if catalogs and Internet, - CS = 1 if catalogs with stores, - IS = 1 if Internet and stores, - CIS = 1 if catalogs, Internet and stores
o DAgh = 1, when a customer h is brought into by a door agent o cs_hath = the number of catalogs sent o ξ = the random error term
Multiple Regression model to test the following hypothesis:
o Customer revenues depend on the channel combination which is used by the customer
oand is higher for the multiple channel combinations
Models elaboration
4
hhhh
hthththththtththt
AgeGenDAgREVISCSCIOSOIOCRET
ξαααα
αααααααα
+++++
+++++++=
1110ht98
7654321
cs_hat
o Gen = Gender of customer h o Age is the age of the customer
o ξ = the random error term
Logit model to test the following hypothesis:
o Customer future profitability is determined by the channel combination which is used by the customer
oand the retention is higher for the multiple channel combinations
The dataset: Customer Channel Behavior
5
2002 2006
Sample size
Percentage of the
population
Sample size
Percentage of the
population
Only Catalogs 1539 34% 360 38%
Only Internet 82 2% 58 6%
Only Stores 2400 53% 504 53%
Catalogs+Internet 143 3% 17 2%
Catalogs+Stores 285 6% 10 1%
Internet+Stores 42 1% 2 0% Catalogs+Internet+Stores 17 0% 0 0%
Total 4508 100% 951 100%
One-channel 4100 89% 4558 99,4%
Multichannel 488 11% 30 0,6%
o Traditional channels used the most and separately
o Multichannel customers constitute only 11%
o Internet gained share
Spending levels and descriptive statistics
6
Mean Std Dev Min Max Median
C €96 (93) €7 €1.496 €68
I €86 (80) €5 €834 €64
S €72 (53) €0,2 €1.243 €60
CI €90 (57) €25 €430 €72
CS €82 (71) €20 €1.211 €67
IS €73 (44) €21 €259 €60
CIS €89 (38) €40 €207 €83
Gender F 70%
M 30%
Average age 43
Catalogs received (per year, median)
4,0
E-mails received (per year, median)
0,7
Door Agent 48%
Street Agent 52%
o Sample: Women, 43 years old
Total euro spent over the relationship
o CIS combination shows the highest average annual spending (median) and lowest dispersion
Channel shares and retention rates
7
Clients channel choices by year, channels share percentage
Retention rates distribution
The channel of future purchases depends on the channel of customer acquisition
8
Channel used
Channel of acquisition
in 2002 Street Agent
% Door Agent % Total
Only C 283 12% 1256 59% 1539 Only I 33 1% 49 2% 82 Only S 1859 79% 541 25% 2400 CI 28 1% 115 5% 143 CS 127 5% 158 7% 285 IS 24 1% 18 1% 42 CIS 7 0% 10 0% 17 Total
2384
100%
2204
100%
4588
Door Agents sign in Catalogs users, Street Agents bring in Store-loyal consumers
Moreover, this relationship remains constant as shows the same Chi square test for each of the following years
The results of the models estimations
9
The customers who were signed into by door agents more likely demonstrated multichannel pattern and higher spending. However, there were not the strong causal relationships in the data identified by the suggested regression model.
Customers’ profitability in the short run turns out to be higher for the Internet only customers. Multichannel combinations generate just slightly more revenues for the company compared to only Store or Catalogs options.
Analysis revealed that in the first year the probability of retention decreased the most, about 9%, everything else being equal, for the two-channel combination Internet and stores. As for the revenues, the Internet channel was also the one with highest retention probability, that was around 5% higher than for the previous case.
The results of the models estimations
10
Regression model coefficients for the channel variables are not constant
Both catalogs send and e-mails led to a growth in the firm’s revenues. Each additional catalogue sent to the customers yearly once in a quarter added 20 euro more in customer spending, given the channel combinations chosen and the channel of acquisition. However, the effect of the e-mails was not profitable, with an estimated – 0.5 euro total effect.
Conclusions
11
The managers should first of all regard the stimulation of usage of more channels by their customers as an instrument to create more purchase occasions, which will generate higher revenues. Yet, depending on the product category the benefits of multichannel distribution will vary. It will be necessary to compare that benefit with the costs it requires, both those of initial investment into the creation of a new channel for the company and the cost of running all the operations in an efficient and integrated manner.
12
Thank You