In this Heavybit Speaker Series, Brian Balfour, VP of Growth at HubSpot, covers the key documents to building a growth process, tactics for generating growth ideas, and what you need to generate a minimum viable test.
4. Brian Balfour :: @bbalfour :: http://www.coelevate.com
GROWING IN A WORLD OF SOFTWARE
HAS CHANGED…
The Lines Between Marketing and Product Are Blurring
Where does marketing end and product begin in software? 1
5. Brian Balfour :: @bbalfour :: http://www.coelevate.com
GROWING IN A WORLD OF SOFTWARE
HAS CHANGED…
The Lines Between Marketing and Product Are Blurring
Where does marketing end and product begin in software? 1
Marketing Has Become More Technical & Quantitative
“Mad Men” skills aren’t as important. 2
6. Brian Balfour :: @bbalfour :: http://www.coelevate.com
GROWING IN A WORLD OF SOFTWARE
HAS CHANGED…
The Lines Between Marketing and Product Are Blurring
Where does marketing end and product begin in software? 1
Marketing Has Become More Technical & Quantitative
“Mad Men” skills aren’t as important. 2
Data Is Becoming More Accessible
Gathering quantitative and qualitative data is becoming easier 3
7. Brian Balfour :: @bbalfour :: http://www.coelevate.com
GROWING IN A WORLD OF SOFTWARE
HAS CHANGED…
The Lines Between Marketing and Product Are Blurring
Where does marketing end and product begin in software? 1
Marketing Has Become More Technical & Quantitative
“Mad Men” skills aren’t as important. 2
Data Is Becoming More Accessible
Gathering quantitative and qualitative data is becoming easier 3
Scale and Speed Of Growth Is Accelerating
Time to 1M Users/ARR has accelerated. 4
9. GROWTH EFFORTS REQUIRE…
Brian Balfour :: @bbalfour :: http://www.coelevate.com
Product
Engineering
GROWTH
Marketing
Data
10. GROWTH AND PRODUCT IS DIFFERENT
BUILDING
CORE VALUE
GETTING THE LARGEST
% OF OUR AUDIENCE
TO EXPERIENCE CORE
VALUE AS QUICKLY AS
POSSIBLE
GETTING USERS TO
EXPERIENCE THAT
CORE VALUE AS OFTEN
AS POSSIBLE
Brian Balfour :: @bbalfour :: http://www.coelevate.com
11. GROWTH EFFORTS
CREDIT: ANDY JOHNS
0
10000
20000
30000
40000
NATURAL ADOPTION
TRUE GROWTH POTENTIAL
PURPOSE
OF GROWTH
EFFORTS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
12. GROWTH IS MORE ABOUT A CHANGE TO
TEAM STRUCTURE, PEOPLE, METHODOLGY,
PROCESS NOT TACTICS/HACKS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
13. LEARNING HOW TO GROW AUTHENTICALLY
DOESN’T START WITH TACTICS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
30. 24
3. RATE OF CHANGE IS ACCELERATING
Brian Balfour :: @bbalfour :: http://www.coelevate.com
31. SEO No Keyword Data, Author Rank, Content
Freshness, Social Shares, Search Result Format
DISPLAY RTB, No Cookies, Mobile
EMAIL Tabbed Inboxes, Inbox Actions, Unsubscribe,
Mobile
FACEBOOK
Newsfeed Ads (4 formats), Video Ads,
Optimized CPM, CTA Format, Custom Audiences,
etc, etc
37. THE GOALS OF OUR PROCESS
1
2
LEARNING
Constant learning of our customer, product, channels
and feeding that learning into the process to improve.
3 AUTONOMY Individuals decide what they work on to achieve the
OKR’s.
4 ACCOUNTABILITY
With autonomy, comes accountability. You
don’t have to be right all the time, but there
is an expectation to improve.
RYTHYM
Momentum is powerful. Establish a cadence to fight
through failures, get to successes, and find
momentum
Brian Balfour :: @bbalfour :: http://www.coelevate.com
38. 1. Brainstorm
(Backlog)
2. Prioritize
(Experiment Doc + Backlog)
4. Implement
5. Analyze
(Experiment Doc)
3. Test
(Experiment Doc)
6. Systemize
(Playbooks)
Set Goals
OKR’s
Explore Data
Qualitative and Quantitative
Find Levers
Growth Model
DAILY/WEEKLY
39. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS TO FIND THE HIGHEST IMPACT AREA THAT WE
CAN FOCUS ON RIGHT NOW GIVEN LIMITED
RESOURCES.
Brian Balfour :: @bbalfour :: http://www.coelevate.com
42. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS WAU’s, BUT WHAT EFFECTS WAU’s?
Brian Balfour :: @bbalfour :: http://www.coelevate.com
43. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS WAU’s, BUT WHAT EFFECTS WAU’s?
WAU’s = New Activated + Retained
Brian Balfour :: @bbalfour :: http://www.coelevate.com
44. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS WAU’s, BUT WHAT EFFECTS WAU’s?
WAU’s = New Activated + Retained
New Activated = (RegisteredFB ADS * ARFB ADS) + (RegisteredViral * ARViral)+…
Brian Balfour :: @bbalfour :: http://www.coelevate.com
45. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS WAU’s, BUT WHAT EFFECTS WAU’s?
WAU’s = New Activated + Retained
New Activated = (RegisteredFB ADS * ARFB ADS) + (RegisteredViral * ARViral)+…
RegisteredViral = Imp * Conversion Rate * Invites/User * Email CTR * Landing Page CR
Brian Balfour :: @bbalfour :: http://www.coelevate.com
46. FINDING LEVERS
BUILDING A GROWTH MODEL
GOAL IS WAU’s, BUT WHAT EFFECTS WAU’s?
WAU’s = New Activated + Retained
New Activated = (RegisteredFB ADS * ARFB ADS) + (RegisteredViral * ARViral)+…
RegisteredViral = Imp * Conversion Rate * Invites/User * Email CTR * Landing Page CR
Retained = (New ActW-1 * RetentionW-1) + (New ActW-2 * RetentionW-2)…
Brian Balfour :: @bbalfour :: http://www.coelevate.com
48. 1. Brainstorm
(Backlog)
2. Prioritize
(Experiment Doc + Backlog)
4. Implement
5. Analyze
(Experiment Doc)
3. Test
(Experiment Doc)
6. Systemize
(Playbooks)
Set Goals
OKR’s
Explore Data
Qualitative and Quantitative
Find Levers
Growth Model
DAILY/WEEKLY
49. HOW WE SET GOALS (OKRs)
OBJECTIVE: Qualitative statement
TIMEFRAME: 30 - 90 Days
KR1: Measurable Goal 1 (Hit 90% Of Time)
KR2: Measurable Goal 2 (Hit 50% Of Time)
KR3: Measurable Goal 3 (Hit 10% Of Time)
Brian Balfour :: @bbalfour :: http://www.coelevate.com
50. AN EXAMPLE…
OBJECTIVE: MAKE VIRALITY A MEANINGFUL
CHANNEL
TIMEFRAME: 90 Days
KR1: 1.5% Viral/WAU
KR2: 2% Viral/WAU
KR3: 3% Viral/WAU
Brian Balfour :: @bbalfour :: http://www.coelevate.com
51. 1
2 IN OUR FACE
INPUTS NOT OUTPUTS
3 CONCRETE
OKR’s will never be set on on the
output (WAU) the will always be
set on the inputs.
OKR’s will be somewhere that we can’t forget
them. Review daily and weekly.
Goals will change mid-OKR VERY rarely.
OKR PRINCIPLES
52. 1. Brainstorm
(Backlog)
2. Prioritize
(Experiment Doc + Backlog)
4. Implement
5. Analyze
(Experiment Doc)
3. Test
(Experiment Doc)
6. Systemize
(Playbooks)
Set Goals
OKR’s
Explore Data
Qualitative and Quantitative
Find Levers
Growth Model
DAILY/WEEKLY
53. HOW WE GET GROWTH INSIGHTS
INSIGHT
Quantitative Qualitative
Intuition
54. 1
2 QUANTITATIVE
QUALITATIVE User Surveys, 1 on 1 Conversations, Full Story
Videos, Customer Support Tickets
GA, Mixpanel, Datastore. Goal is to answer
questions we don’t know about our user, product,
or channel. Break it into pieces.
EXPLORE DATA
UNDERSTAND THE INPUTS OF THE INPUTS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
55. BUILD A MINI MODEL
UNDERSTAND THE INPUTS OF THE INPUTS
1
2 CEILING
BASELINE
3 SENSITIVITY
56. UNDERSTAND THE WHY
UNDERSTAND THE INPUTS OF THE INPUTS
Why do users who get to invite page not covert?
<— Super targeted to the variable we are trying to improve
<— Open ended question to allow for all types of answers
<— Email looks like a personal email to create a sense of
importance
57. UNDERSTAND THE WHY
UNDERSTAND THE INPUTS OF THE INPUTS
Bug
7%
Other
8%
Would Prefer To Just Share On FB
10%
Don't Want You To Spam Friends
11%
Reward Not High Enough
29%
Didn't Know Who To Invite
35%
60. STEP ONE: BRAINSTORM
ACTIVATION RATE = STEP 1 CONV * STEP 2 CONV * STEP 3 CONV
IF WE WANT TO
IMPROVE THIS
WE BRAINSTORM ON THESE
Brian Balfour :: @bbalfour :: http://www.coelevate.com
61. 4 WAYS TO COME UP
WITH GROWTH IDEAS
1 OBSERVE How are others doing it? Look outside of your
immediate product space. Walk through together.
2 QUESTION Question brainstorming. Why? What is… What if…
What about… How do we do more of…
3 ASSOCIATE Connect the dots between unrelated things. i.e What
if our activation process was like closing a deal?
4 NETWORK Growth mastermind groups. Build your network of
growth people. I’m on the phone every day for ideas
Brian Balfour :: @bbalfour :: http://www.coelevate.com
63. STEP TWO: PRIORITIZE
1
2 IMPACT
PROBABILITY
3 RESOURCES
LOW - 20%, MEDIUM - 50%, HIGH - 80%
Comes from your prediction. Take into account
long lasting effects vs one hit wonders.
Marketing, Design, Engineering
64. STEP TWO: PRIORITIZE
CREATING A HYPOTHESIS
IF SUCCESSFUL, [VARIABLE] WILL
INCREASE BY [IMPACT], BECAUSE
[ASSUMPTIONS].
Brian Balfour :: @bbalfour :: http://www.coelevate.com
65. STEP TWO: PRIORITIZE
JUSTIFYING YOUR ASSUMPTIONS
1
2 QUALITATIVE
QUANTITATIVE
3 SECONDARY
Previous Experiments, Surrounding Data,
Funnel Data
Surveys, Support Emails, User Testing
Recordings
Networking, Blogs, Competitor Observation,
Case Studies
Brian Balfour :: @bbalfour :: http://www.coelevate.com
68. STEP THREE: TEST
WHAT IS THE MVT
(MINIMUM VIABLE TEST)
TO UNDERSTAND OUR HYPOTHESIS?
Brian Balfour :: @bbalfour :: http://www.coelevate.com
69. 1 EFFICIENCY What is the least resource intensive way to gather
data about the hypothesis.
2 VALIDITY
The experiment must take into account how to get
a valid result by designing a control and required
amount of data.
STEP THREE: TEST
Brian Balfour :: @bbalfour :: http://www.coelevate.com
70. STEP THREE: TEST
WRITE THE EXPERIMENT DESIGN IN EXPERIMENT DOC
Brian Balfour :: @bbalfour :: http://www.coelevate.com
74. STEP FIVE: ANALYZE
1
2 IMPACT
SUCCESS/FAIL
3 WHY?
How close to your prediction?
The most important question you can ask. Why did
you see the result that you did?
Be prepared for a lot of failures.
75. STEP FIVE: ANALYZE
LEARNINGS + ACTION ITEMS IN EXPERIMENT DOC
Brian Balfour :: @bbalfour :: http://www.coelevate.com
77. STEP SIX: SYSTEMIZE
1 PRODUCTIZE Productize as much as you can with technology
and engineering.
2 PLAYBOOKS For the things you can’t productize, build into step
by step playbooks to make them repeatable.
Brian Balfour :: @bbalfour :: http://www.coelevate.com
79. WHAT A TYPICAL WEEK LOOKS LIKE
Monday Tuesday Wednesday Thursday Friday
Growth Meeting
- Learnings
- Goal Review
- Blocking Items/
Strategies
Predicting, Prioritizing, Implementing,
Analyzing, Systemizing
Brian Balfour :: @bbalfour :: http://www.coelevate.com
80. WEEKLY GROWTH MEETING
DON’T GET ATTACHED TO IDEAS
1
2 GOALS
LEARNINGS
3 BLOCKING
Learnings, not updates. You can monitor
experiment updates in the pipeline.
Progress against goals
What blocked you from doing something this
week?
82. MACRO OPTIMIZATION
(REVIEW QUARTERLY)
1
2 ACCURACY
BATTING AVG
3 THROUGHPUT
How many successes to failures? Our we
improving over time?
Are our predictions getting more accurate?
How many experiments are we running in a given
time period? How do we do more?
83. MACRO OPTIMIZATION
HOW WE IMPROVE
1
2 INFRASTRUCTURE
TEAM
3 PROCESS
Improve current teams skills. Add to the team.
Experiment, analytics, and qualitative
feedback infrastructure
Refining the process for more efficiency or
quality of ideas, implementation, and analysis.
84. A FINAL WORD PART ONE
ITERATION DOES NOT EQUAL
INCREMENTAL
Brian Balfour :: @bbalfour :: http://www.coelevate.com
85. A FINAL WORD PART TWO
IT IS NEVER TOO EARLY
TO START THIS PROCESS
Brian Balfour :: @bbalfour :: http://www.coelevate.com
86. A FINAL FINAL WORD
THERE IS NOTHING SPECIAL ABOUT THIS
PROCESS. SUCCESS COMES DOWN TO
GRIT, FOCUS, AND PERSISTENCE.
Brian Balfour :: @bbalfour :: http://www.coelevate.com