2004 boston globe
DESCRIPTION
TRANSCRIPT
Retention Marketing
At the
Boston Globe
2
Retention Marketing
• Get familiar with Retention Marketing (RM) at the Boston Globe
• RM campaigns
• Future RM projects
Purpose
Agenda • Formation of Relationship Marketing Group (RMG)
• Current RMG Campaigns
credit card conversion campaign
sequencing campaign
prepaid campaign
extend discount terms campaign
• Customer Attrition Model
• Future Work of RMG
• Q&A
Purpose and agenda
3
Agenda
Customer Attrition Model
Current RMG Campaigns
Formation of RMG
Future Work of RMG
Q&A
4
Formation of RMG – A Blended Call Center/Customer Service Environment
Four Basic Principles of RMG
Current RMG Campaigns:
Stop Saves, Upgrade, Credit Card Conversion, Overdue/Non-pay, Winback/Expired calling/mailing, New Start welcome calling/mailing, Globe Rewards etc.
• Nurture a lasting relationship
• Convert traditional payment by bill to auto-payment or Prepay
• Prevent stops before they happen
• Reward loyal customer
5
Agenda
Current RMG Campaigns
Formation of RMG
Future Work of RMG
Customer Attrition Model
Q&A
6
Current RMG Campaigns - Credit Card Conversion
• Overview
• 37% of new starts from Telemarketing come with credit
card payment monthly
• 50% active subs pay by credit card
%
Active subs
pay by credit
card 50.48%
Active subs
pay by bills 49.52%
• Why are we interested in credit card payment ?
• one-year retention rate is 43.8% higher for new starts
from Telemarketing channel who pay by credit card vs.
those do not
• Credit Card Conversion Model
• Simple model did in Analytix to identify potential variables that drives credit card conversion
• Get basic profile of credit card payers; applied the profile to non-credit card payers
One year TM
Retention Rate
New Starts with
credit card
57.41%
New Starts without
credit card
32.88%
7
Current RMG Campaigns - Credit Card Conversion (Cont’d)
• Basic profile of active subscribers who pay by credit card
Internet Usage % of active w ith
credit card
1 49.13%
2 14.02%
3 9.37%
4 6.00%
5 4.74%
6 4.21%
7 3.98%
8 4.15%
9 3.07%
10 1.33%
11 0.01%
Total 100.00%
Tenure % of active w ith
credit card
0 to 6 Years 60.11%
6 to 10 Years 9.61%
10 Years More 30.28%
Total 100.00%
Household head
Age
% of active w ith
credit card
0 TO 24 2.18%
25 TO 29 2.38%
30 TO 34 5.93%
35 TO 39 9.71%
40 TO 44 12.66%
45 TO 49 12.76%
50 TO 54 12.58%
55 TO 59 10.82%
60 TO 64 7.75%
65 TO 69 5.97%
70 TO 74 5.94%
75+ 11.32%
Total 100.00%
Household Income % of active w ith
credit card
Less than 35000 12.24%
35000 to 70000 21.40%
70000 + 66.35%
Total 100.00%
Delivery Maps % of active w ith
credit card
7 Day
73.88%
Sunday Only 21.37%
Others 4.75%
Total 100.00%
8
Current RMG Campaigns - Credit Card Conversion (Cont’d)
• What we learned from basic profiling ?
• Subscribers who are intensive internet users are more likely to use auto-payment
• Subscribers who have tenure between 0 to 6 years are more likely to use auto-payment
• Subscribers who are aged between 40 to 59 years old are more likely to use auto-payment
• Subscribers whose household income is greater than $70K are more likely to use auto-pay
• Subscribers who are on 7-day delivery map are more likely to use auto - pay
• Segmentation
9
Current RMG Campaigns - Credit Card Conversion (Cont’d)
• Apply profile of auto-pay subscribers to non-auto payers
• RMG called two files on 7-day and Sunday only subscribers
• Sales/contact rate increased by over 3 times compared to that from old segmentations, which basically called everyone who is not auto-payers
• Results
• Another successful story from Telemarketing acquisition on Credit Card Payment
• Patriot T-shirt Offer awarded new starts from TM acquisition paying by credit card with a
free T-shirt
• Generate a 22% higher acquisition rate of credit card payment than giving 4 additional
weeks of free subscription
Maps Sales Per Hour % of Contact % sales/contact rate
Old
Segmentation All Maps 0.33 19.69 1.94
7 Day 1.52 21.4 8.08
Sunday Only 1.55 39.81 6.4
New
Segmentation
10
Current RMG Campaigns - Credit Card Conversion (Cont’d)
• Conclusions
• Credit card payment is critical to TM acquisition and retention
• Simple modeling by subscribers’ profile will help increase credit card conversion rate
• Premium offer generates higher credit card conversion than does longer discounts periods or free
additional weeks
11
Current RMG Campaigns – Sequencing
Winback Direct Mail/Calling Sequencing
12
Current RMG Campaigns – Email Campaign
• Overview
• The Globe has 300,000 + email address,
• 100,000 subscribers’ email address
• 44,000 opt in subscribers
• 82% of opt-in with email addresses
• 72% are 7 day subscribers
• Major Source of Emails
• Boston.com
• The Globe Services
• The Boston Globe Rewards
• PBS (Circulation system)
13
Current RMG Campaigns – Email Campaign (Cont’d)
• Conclusions
• Email campaign costs much less than direct mailing or telemarketing
• Email campaign generates much better upgrade rate than tradition direct mailing
• First Globe Upgrade Email Campaign
Delivery Maps Total Mailed Net Delivered Delivery Rate Open Rate Total Click Rate Unsub Rate Sales % of sales Cost Per Order % of sales Cost Per Order
Sunday 9,540 7,989 83.70% 32.80% 7.70% 1.40%
Thursday-Sunday 1,369 1,138 83.10% 30.80% 11.80% 0.80%
Other 845 703 83.20% 32.80% 7.00% 1.60%
Total 11,754 9,830 83.60% 32.60% 8.10% 1.30% 166 1.69% $8.00 0.31% $86.50
Upgrade Email Campaign Direct Mail Upgrade Campaign
14
Current RMG Campaigns – Prepaid Campaign
• Objective
• Bring more subscribers to long-term payment plans for 13, 26, 52 week terms
• Build long-term customer retention
• Profile of Prepaid subscribers
• 77% of prepaid subscribers are on 52-week term; 22% are on 26-week term
• 88% prepaid subscribers have tenure of over 2 years
• 80% prepaid subscribers are on 7-day map
• Majority of prepaid subscribers are seniors
• Methodology
• Direct mailing or RMG calling to active subscribers for prepayment
• Results
• 1.1% response rate of converting to pre-payment out of 1.5 million direct mailing, RMG
calling and inserts
15
Current RMG Campaigns – Prepaid Campaign (cont’d)
• Retention of Prepayment subscribers
Tenure Payment Type Subscribers Attritors % Attrition Subscribers Attrition % Attrition
0-6 Months
Non-Credit Card 25 - 0.0% 156 87 55.8%
Credit Card 74 - 0.0% 42 - 0.0%
Sub-Total 99 - 0.0% 198 87 43.9%
7-12 Months
Non-Credit Card 58 - 0.0% 325 179 55.1%
Credit Card 162 2 1.2% 113 - 0.0%
Sub-Total 220 2 0.9% 438 179 40.9%
13-24 Months
Non-Credit Card 164 5 3.0% 707 260 36.8%
Credit Card 415 8 1.9% 457 1 0.2%
Sub-Total 579 13 2.2% 1,164 261 22.4%
25+ Months
Non-Credit Card 3,817 19 0.5% 13,908 1,923 13.8%
Credit Card 8,323 40 0.5% 10,596 9 0.1%
Sub-Total 12,140 59 0.5% 24,504 1,932 7.9%
All Tenure Totals
Non-Credit Card 4,064 24 0.6% 15,096 2,449 16.2%
Credit Card 8,974 50 0.6% 11,208 10 0.1%
Grand Total 13,038 74 0.6% 26,304 2,459 9.3%
Total Total
Pre-Paid Subscribers Control Group
• Conclusions
• Overall attrition rate for prepay subscribers is 0.6% compared to that of 9.3% for non-prepay group
• For credit card payers, attrition rate of non-prepay group is less than that of pre-pay subscribers
• It is more important to get subscribers auto-pay by credit card than to get them pre-pay
16
Current RMG Campaigns – Discount Extension Campaign (Extendo)
• Objective
• Increase retention rate on new subscribers by extending their initial discount terms
• Methodology
• Extending initial discount offer period to longer terms
• Target at subscribers on 25%, 50%, 75% off regular rates, and extend their offer terms
• Results
Tenure Extended Attrition Extended Attrition
0 to 6 months 3,151 8.73% 135 22.96%
7 to 12 months 2,405 9.73% 259 24.32%
13 to 24 months 1,868 9.15% 3,205 27.64%
25 months + 8,256 9.04% 7,337 19.46%
Total 15,680 9.09% 10,936 22.02%
Extendo Group Control Group
(Results were calculated after two month of campaign being launched)
17
Current RMG Campaigns – Extendo Campaign (Cont’d)
• Conclusions
• Average two-month attrition rate for Extendo campaign is 9.1% vs. 22% for Control group
• Retention is 2.5 times favorable for Extendo compared to that of Control group
• Retention is most favorable for new subscribers with tenure of less than 6 months
18
Agenda
Current RMG Campaigns
Customer Attrition Model
Formation of RMG
Future Work of RMG
Q&A
19
Subscribers
Customer Attrition Model - Customer Dynamics at The Globe
Prospects
DM TM
250 k / year
240 k / year
Developments with TM legislation further increases importance of retention
Acquisition and Retention are VERY leveraged improvement opportunities
20
Customer Attrition Model – Objective
What are we trying to predict
10 2 3 4 5 610 2 3 4 5 6
Probability of attrition during
a 6-month window
Time of scoring
Independent of:
What the install channel was
How long they’ve been a customer
What product they currently have
What discount they are on
21
Customer Attrition Model – Overview of Variables
Customer Attrition Model
Defining the Dependant Variable
• The dependent variable is defined as the probability that a given customer will attrite sometime during the next 6-month window
Data Used
• Data sources: Customer Database – (Subscriber, Transaction, Suppression), TM Workstation (History, Suppression) Donnelley Marketing
Definition of Attrition
• Non-voluntary attrition excluded – excluded stop reasons of Deceased, Hospitalized, Non-payment, etc…
• Sample of 50,000 used for modeling
Technique • Logistic regression is the technique due to the discrete choice nature of the model
22
Customer Attrition Model – Drivers of Attrition
Summary of final model variables
Lower Likelihood Factors Higher Likelihood Factors
• Log of Total number of payments
• Likelihood of teenager in household
• NY Times Subscriber
• Most recent TM campaign was an Upgrade
offer
• Home Value - $175K+
• Number of times customer has been on a STOP TM
campaign
• Canceled for Billing, Service, or Proactive reason (No
time, prefer other paper, etc…)
• Internet Usage Likelihood
• Number of times TM Home Delivery campaign and
the result was a SALE
• Number of times TM disposition Refusal – Hung Up
• Canceled at end of promotion
• Number of times TM disposition Refusal – Reads
Competition
23
Customer Attrition Model – Model Performance
Lift chart for final model
Decile
Sample Predicted Actual Predicted Actual In Decile Cumulative Predicted Actual
1 5,000 74.38% 70.22% 3,719 3,511 22.1% 22.1% 234 221
2 5,000 55.89% 56.78% 2,794 2,839 17.9% 40.0% 176 179
3 5,000 46.71% 47.64% 2,336 2,382 15.0% 55.0% 147 150
4 5,024 38.91% 40.59% 1,955 2,039 12.8% 67.8% 123 128
5 4,976 32.11% 33.58% 1,598 1,671 10.5% 78.4% 101 106
6 5,005 24.02% 27.37% 1,202 1,370 8.6% 87.0% 76 86
7 4,995 17.30% 18.50% 864 924 5.8% 92.8% 54 58
8 5,000 12.65% 12.02% 632 601 3.8% 96.6% 40 38
9 5,000 9.33% 6.62% 466 331 2.1% 98.7% 29 21
10 5,000 6.17% 4.14% 308 207 1.3% 100.0% 19 13
50,000 31.75% 31.75% 15,875 15,875 100.0% 100.0%
Top to Bottom ratio = 17.0
Attrition Index
LIFT CHART FOR CUSTOMER ATTRITION MODEL
% That Attrited # That Attrited % of All Attritors
24
Customer Attrition Model – Model Performance (Cont’d)
Gains chart for final model
LIFT CHART FOR CUSTOMER ATTRITION MODEL
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
1 2 3 4 5 6 7 8 9 10
Decile
% A
ttri
tors
in
Dec
ile
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
Cu
mu
lati
ve %
of
all
Att
rito
rs
Predicted Actual Cumulative
78% of all attritors
captured in top half
of the data
55% of all attritors capured
in the top 3 deciles
25
Customer Attrition Model – What Now?
Uses for the Customer Attrition Model
• Identify churner or discount junkies
• Identify loyal customers who are most likely to attrite
• Identify subscribers who are on full-price rate and are most likely to attrite
• Identify non-pay subscribers who are most likely to attrite
• Identify all other “at-risk” subscribers
• Target Retention Programs
Campaigns for the Above Targets?
• Get more on credit card payment or long-term pre-payment
• Extend discount offers
• Differentiated offers to different targets
• Final bill notice
• Overdue and no-pay calls
26
Customer Attrition Model – An Example of NonPay Stop
• Increase retention rate by prevent non-payment stops
• Objective
• Convert subscribers who have multiple non-pay histories to prepayment or credit card payment
• Overdue/non-pay calling
• Final bill notice by invoices or in paper
Last Stop Reasons Percentage
Vacation stop 20.41%
No time to read 13.85%
Non Payment stop 11.39%
Moving 9.29%
Miscellaneous 7.31%
Vacation stop with no restart 5.36%
Vacation NIE donation 5.12%
Never Ordered 3.78%
Customer service 3.38%
Offer Over 1.91%• Methodology
• Results
Mailing final notice is the most effective to have subscribers
to pay overdue or non-pay balance Paid upon final notice
Mail 75.90%
Carrier 40.00%
Calling 12.00%
27
Customer Attrition Model – An Example of NonPay Stop (Cont’d)
• Non-pay Subscribers by Customer Attrition Deciles
• Top 5 customer attrition deciles captured over 72% non-pay stop subscribers
• Top 5 attrition deciles captured 62% of all attritors with small amount of balance due
at non-payment stop
• Conclusions
Customer Attrition
Decile\ Balance
>=$0.00
and
<$10.00
>=$10.00
and
<$20.00
>=$20.00
and
<$30.00
>=$30.00
and
<$40.00
>=$40.00
and
<$50.00
>=$50.00
and
<$75.00
>=$75.00
and
<$100.00
>=$100.00
and
<$150.00
>=$150.00
01 2.03% 0.02% 0.01% 0.00% 0.00% 0.00% 0.00% 0.00% 0.54%
02 7.41% 0.05% 0.03% 0.01% 0.00% 0.01% 0.00% 0.01% 1.27%
03 15.13% 0.11% 0.04% 0.02% 0.01% 0.01% 0.00% 0.01% 1.69%
04 20.20% 0.17% 0.06% 0.01% 0.02% 0.02% 0.01% 0.01% 2.24%
05 17.65% 0.19% 0.07% 0.02% 0.02% 0.03% 0.01% 0.02% 2.60%
06 9.46% 0.23% 0.07% 0.02% 0.02% 0.02% 0.02% 0.01% 2.44%
07 4.94% 0.27% 0.10% 0.02% 0.02% 0.03% 0.02% 0.02% 2.13%
08 2.29% 0.21% 0.09% 0.02% 0.01% 0.03% 0.01% 0.01% 1.66%
09 0.96% 0.18% 0.10% 0.02% 0.01% 0.02% 0.00% 0.01% 0.91%
10 0.80% 0.13% 0.06% 0.01% 0.01% 0.01% 0.01% 0.01% 0.87%
28
Agenda
Current RMG Campaigns
Future Work of RMG
Formation of RMG
Customer Attrition Model
Q&A
29
Future Work of RMG
• Predictive model of credit card conversion
• Identify variables that drives credit card conversion
• Predict the likelihood of non-credit card paying subscribers converting to credit card payment
• Increase overall retention by getting more subscribers/prospects to credit card payment
• Expired Model
• Predict the likelihood of former subscribers to come back
• Identify variables that most drive expires to come back
• Identify differentiated offers that most likely to get expires to come back
• Customer Segmentation
• Better understand Globe’s subscribers’ profile
• Target campaigns at the right group of audience
• Preliminary thoughts : new subscribers, chronic churners, loyal subscribers, seniors,
seasonal subscribers…
• Upgrade Model
• Predict the likelihood of subscribers to be upgraded
• Identify variables that most drive upgrade
• Identify differentiated offers that most likely to upgrade subscribers
30
Future Work of RMG – Summary
Potential next steps for the Boston Globe
Channel
Optimization
Market to prospects
with a combination
of likelihood to
respond AND retain
DM / CRM Capabilities
Customer Value
Management
LTV Optimization
Predictive Models
Market to likely
responders, based
on channel.
Independent
decisions
Offer Management
Build on targeting
optimization with offer
management. Offer
discounts where
needed and
appropriate
Cross-Channel
Optimization
Make cross-channel
decisions based on
likelihood to
respond, stay, and
cost per sale
Customer Attrition model
can impact decisions for
Offer Management, CVM,
and ultimately, LTV
Le
ve
l o
f S
op
his
ticatio
n a
nd
im
pa
ct
Low
High
High
Customer Attrition model
can impact decisions for
Offer Management, CVM,
and ultimately, LTV
31
Agenda
Credit Card Conversion
Q & A
Formation of RMG
Future Work of RMG
Customer Attrition Model