2. Eric Kirby Hayley Osher
CEO Director, Member Engagement
ekirby@connectionengine.com hayley@hautelook.com
Twitter: @ConnectionEng
3. Inactive Subscribers – Why care?
• Often 50%+ of a subscriber file Why Inactive?
Not “in market”
• Not mailing causes issues Passive unsub
– Potential revenue impact Delivery issues
– Lost future contact opportunity Seldom used account
• Mailing causes issues. Pressure to suppress:
– Deliverability concerns
– Customer satisfaction
– Desire to reduce marketing waste
4. What are your options?
• Attempt to reactivate (Messaging, Offers)
• Reduce mail cadence
• Suppress
• How will you decide who to suppress?
– Who has the potential to become a valuable customer?
– Who is likely to be worth little no matter how many times mail?
What if you knew the future value
potential among the inactives?
5. How?
• Determine “Value Potential” for each subscriber by
leveraging external data and analytics.
• Process:
– Quantify attributes of Best Customers.
– Develop a mathematical model to predict customer value.
• Based on how much subscriber “looks like” a Best Customer.
– Use model to segment subscribers by Value Potential.
• Fast and cost effective through SaaS application
6.
7. Case Study
• Leading flash sales site. Acquired by Nordstrom.
• Sale events each day featuring best names in fashion,
beauty, kids, and home.
• Large and rapidly growing member base.
8. Segmentation Strategy
• Sophisticated approach to segment users by observed
purchase behavior and email engagement.
– 38 defined member lifecycle segments defined by time on file,
engagement, and purchase activity.
• Aggressively suppress inactives to minimize
deliverability issues.
• How to differentiate among inactives?
– By definition, little available data to segment
9. Step 1: Profile High Value Customers
• High Value Members have distinctive attributes
compared to Members who have never purchased.
• Examples:
High Value vs.
Connection Engine Attribute Non Buyer
Home Value 2.0 x
Online Purchase Activity 2.8 x
High Fashion Apparel Buyer 3.5 x
"Young & Hip" Lifestyle 2.6 x
Amex Cardholder 3.3 x
10. Step 2: Segment by ―Value Potential‖
• Using external data attributes, mathematical models are
used to predict individual member “Value Potential”.
Value Potential
Segment % of Members Relative Value
1 - Highest 10% 16 x
2 20% 8x
3 20% 4x
4 20% 2x
5 - Lowest 30% 1x
11. Step 3: Programs by Segment
• New members segmented daily.
Value Potential Segment
Source 1 2 3 4 5
A 12% 16% 35% 31% 7%
B 6% 27% 29% 26% 12%
C 3% 13% 17% 33% 35%
• Can identify new subscribers who have the potential to
be high value from the moment they subscribe.
– Enables tailored communication / offer strategy
• Inactive members: reactivation campaign implemented
and measured by Value Potential segment.
13. Relative Performance by Segment
One Month after Reactivation Message
Low Segment High Segment
4.4
3.1
2.2
1.8
1 1 1 1
Open Rate Login Rate AOV Order Rate
14. Revenue per Member 14x for High Segment
• Revenue per Member Low Segment High Segment
14x higher in High
segment vs. Low. 14
• High Segment members
were 60% less likely to
redeem coupon when
purchasing vs. Low!
• High Segment inactives 1
represent ideal long term
customers to re-engage. Revenue per Member
15. Conclusions
• High Value Potential segment members represent best
of the inactive universe.
– Investments to reengage this segment
– Worth keeping on file even after long periods of engagement
• Low Potential inactives represent lowest value members.
– Unlikely to ever become valuable customers
– Safest universe to suppress
• Value Potential segmentation provides a powerful tool to
determine who to keep on file, and who to suppress.
16. Questions?
Eric Kirby Hayley Osher
CEO Director, Member Engagement
ekirby@connectionengine.com hayley@hautelook.com
Twitter: @ConnectionEng
19. SQUEEZING BLOOD FROM A
STONE?
According to lead judge (for MarketingSherpa Email Awards 2012) Adam T.
Sutton, Senior Reporter, MECLABS Primary Research, "Email marketers are
constantly challenged to extract more value from their lists. Many turn to
inactive or non-purchasing subscribers to boost conversions – but this can
feel like squeezing blood from a stone.”*
*http://www.strongmail.com/resources/blogs/email-marketing-insights/2011/11/travelocity-win-back-campaign-wins-marketingsherpa-email-awards-2012
20. Case Study—Workopolis
• Background—Workopolis is the leading site in Canada who’s
goal is to connect Seekers and Employers
– We connect the right Seekers with the right Employers, across the country,
across a multitude of job categories
• Have two types of lists, B2C and B2B
• B2B is where money is made by charging for job postings
• B2C has no money associated with it but they need to keep users engaged
– Currently developing full life-cycle programs for both sides of
business
– Don’t worry much about invalid account, focus on engagement
with ISP’s
• Look at recent activity
21. Job Seeker (B2C) Model
• 4 key job seeker behaviors
– Visitor Career Alert Creation
– Register
– Upload a Resume
– Apply
• Certain key behaviors leads to an email communicating the importance of
“taking the next step”
– Time relevancy = increased conversion
Acquisition Growth Win back
Visitor Upload Modify
Join Apply Lapsed
Career a Your
Now Now Users
Alert Resume Resume
Convert Convert Convert
22. Targeting Workopolis’ Lapsed Users
• What is a lapsed user to Workopolis?
Win back
– A user who was actively using Workopolis
• Have a resume uploaded to the site
– Based on last login date
• When do we plan to speak to these
Lapsed Users lapsed users?
– Secondary research suggests that job seeker
start looking for a new job after approximately
one year in their current role*
– Site behavior of the average job seeker
• What do we plan on saying to them?
– “Thinking of finding a new job?”
– We want to “Warm them up” for a return to
Workopolis.com *Source: Job Seeker Study, 2010
23. First Steps to Building a Re-
Engagement Program
• Why is it important?
• Setting proper expectations
– Make sure to communicate throughout your organization of
your goals and success measurements
• Determine What an Inactive Customer Means to You
– Leveraging the data you already have
– Don’t try to go back too far
• Examine the Content You Are Sending Them
• Don’t Send Win-Back Emails from the Same System as Your
“Master” List
– Start slow
– Monitor complaints and permanent bounces
24. Recommendations for Maximizing CTR from Openers
• Treat non-clicking openers as a separate segment
– Use clicks among all openers to identify members
– Include non-clickers and infrequent clickers (clicked on <25% of
all opened emails) as segments
• Test different approaches to both segments
– Calls to action
– Copy & Images
• Test resends on a limited basis to prevent burnout
24
25. Things to Remember
• Monitor your mailings closely
• Don’t try to do to much at first, start slowly and
build
• Don’t try to over communicate
• It is ok to say Good-Bye