the holy grail of traction - brian balfour, hubspot
TRANSCRIPT
DISCLAIMER: ALL DATA IN THIS PRESENTATION IS FAKE
(don’t go public if you want to present real data)
Brian Balfour :: @bbalfour :: http://www.coelevate.com
WHAT SEPARATES THE TOP 1% FROM THE BOTTOM 99%
IN GROWTH?
Brian Balfour :: @bbalfour :: http://www.coelevate.com
Brian Balfour :: @bbalfour :: http://www.coelevate.com
ANSWER…RETENTION! GRAPH CREDIT: WWW.ANDREWCHEN.CO & QUETTRA
IF YOU HAVE POOR RETENTION NOTHING ELSE MATTERS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
BAD RETENTION GRAPH CREDIT: DAN WOLCHONOK - SR PM @ SIDEKICK
Brian Balfour :: @bbalfour :: http://www.coelevate.com
GOOD RETENTION GRAPH CREDIT: DAN WOLCHONOK - SR PM @ SIDEKICK
Brian Balfour :: @bbalfour :: http://www.coelevate.com
AWESOME RETENTION GRAPH CREDIT: DAN WOLCHONOK - SR PM @ SIDEKICK
Brian Balfour :: @bbalfour :: http://www.coelevate.com
BUT THERE’S MORE… INCREASE RETENTION
AFFORD HIGHER CPA
INCREASE LTV
INCREASE RETENTION
DECREASE eCPA
INCREASE VIRALITY
INCREASE RETENTION
DECREASE PAYBACK PERIOD
INCREASE UPGRADE RATES
Brian Balfour :: @bbalfour :: http://www.coelevate.com
EVERY IMPROVEMENT TO RETENTION ALSO IMPROVES VIRALITY, LTV, AND
PAYBACK PERIOD
Brian Balfour :: @bbalfour :: http://www.coelevate.com
RETENTION COHORT CURVE
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
Brian Balfour :: @bbalfour :: http://www.coelevate.com
WHAT IS GOOD AND WHAT IS BAD?
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
THE MOST IMPORTANT ELEMENT IS THAT IT FLATTENS
GOOD
BAD
Brian Balfour :: @bbalfour :: http://www.coelevate.com
TO UNDERSTAND AND OPTIMIZE RETENTION YOU NEED TO THINK ABOUT
IT IN ACTIONABLE PIECES
Brian Balfour :: @bbalfour :: http://www.coelevate.com
WEEK ONE RETENTION
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
W1 RETENTION
GET USERS TO EXPERIENCE CORE VALUE AS QUICKLY AS POSSIBLE
Brian Balfour :: @bbalfour :: http://www.coelevate.com
IMPROVING W1 RETENTION SHIFTS THE CURVE UP
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
BETTER ONBOARDING
CLEARER MESSAGING
SEGMENTED NUX
Brian Balfour :: @bbalfour :: http://www.coelevate.com
MID TERM RETENTION
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
MID TERM RETENTION
GET USERS TO CREATE HABITS AROUND CORE VALUE (RE-WIRE)
Brian Balfour :: @bbalfour :: http://www.coelevate.com
LONG TERM RETENTION
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
LONG TERM RETENTION
GET USERS TO EXPERIENCE CORE VALUE AS OFTEN AS POSSIBLE
Brian Balfour :: @bbalfour :: http://www.coelevate.com
SLOW/FLATTEN CURVE OFF
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active! FEATURE REFINEMENT
RETENTION HOOKS
PRODUCT QUALITY
RESSURECTION
Brian Balfour :: @bbalfour :: http://www.coelevate.com
SIDEKICK RETENTION CURVE DEC 2014
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
% Active!
TIME!
Problem #2: Not Flat!
Problem #1: !Over time W1 Retention Was Decreasing!
Brian Balfour :: @bbalfour :: http://www.coelevate.com
1 SEGMENTATION Segmented retention curves by different factors: Acquisition source, email client, persona, type of email
HOW WE FIGURED OUT IMPROVEMENTS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
Brian Balfour :: @bbalfour :: http://www.coelevate.com
CORPORATE VS FREEMAIL
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
Corporate Emails! Gmail and Other Freemail!
% Active!
Time!
Higher W1 Retention!
Slower User Churn!
1
2 QUANT INDICATORS
SEGMENTATION Segmented retention curves by different factors: Acquisition source, email client, persona, type of email
Looked for quantitative indicators by comparing data between those who churned and those who retained.
Brian Balfour :: @bbalfour :: http://www.coelevate.com
HOW WE FIGURED OUT IMPROVEMENTS
1
2 QUANT INDICATORS
SEGMENTATION
3 1:1 QUAL DATA
Segmented retention curves by different factors: Acquisition source, email client, persona, type of email
Looked for quantitative indicators by comparing data between those who churned and those who retained.
Extremely targeted 1:1 emails to users asking open ended question.
Brian Balfour :: @bbalfour :: http://www.coelevate.com
HOW WE FIGURED OUT IMPROVEMENTS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
SOME OF OUR FINDINGS
“Didn’t have time to
figure out.”!
Brian Balfour :: @bbalfour :: http://www.coelevate.com
RETENTION CURVES
0%!
25%!
50%!
75%!
100%!
125%!
0! 5! 10! 15! 20!
Older Cohorts! Newer Cohort 1! Newer Cohort 2!
% Users !Still Active!
Time!
Woohoo!!Curves Shift Up In Latest Cohorts!
Woohoo!!Old Cohorts Flatten!
STRATEGIES THAT WORKED FOR US
1 SEPARATING NEW USER EXPERIENCE
2 SEGMENTING BASED ON PERSONA
3 SWITCHING USER CONTEXT PERSONAL -> WORK
4
5
TESTING MEDIUM OF EDUCATION
CONTENT