Distribution via Continuous Improvement Feedback LoopsGrowth Hacking for the Agile Marketing Meetup
Paul Willard, Atlassian CMO • 11/27/2012
• Boeing Engineer• 7 years writing code
• Learned Japanese manufacturing concepts
• Product Management• Customer acquisition products
• A/B testing
Me
Marketing Growth Hacking
Development
• Startups need technology, product, and distribution
• Many fail on distribution• Myth:
If you build it, they will come
• Fact:The best distribution won
Distribution
• Most products do not have a target market of everyone• Nuclear chain reactions &
explosive virality are difficult to achieve without concentration
• Virality should not be THE plan
Unicorns & Viral Supernovas
• Most ideas will not work• Don’t invest too much in
any one idea – no epic fails• Planned failure - fast, cheap
Learn to Fail
• New media, channels, and platforms come and go quickly• Social, mobile, tablet, Pinterest, Groupon
• AOL, Yahoo, MySpace, Display, Palm Pre, BlackBerry, Meebo, hotmail, del.icio.us . . .
• Tech know how and flexibility needed to apply new sources, data, API’s
Change – The Only ConstantBirth and death timeline instead?
The Only Charts You Need
Month
Cus
tom
ers
Life
time
Valu
e ($
)
Acquisition Exponential Cohort Curves
Months as Customer
• Test quick, cheap, and dirty• Released vs polished
• Front end flexibility
• Standardize reporting
• Measure LTV• Conversion is not good enough
• Model it from twitchy data
The Growth Machine
• Acquisition exponential• Eyeballs
• Conversion Funnel
• Never stop what is working, be incremental & relentless
Feedback Loops
• Prioritize releasing• Cadence - 2 / person week
• Drop things that don’t work
• Put more into things that do
• 2 Backlogs• New eyeballs to product
• Conversion A/B testing
Kanban Team
• 12 month break even• Conservative
• Estimating is ok
• Virality
• Model LTV for each channel or even campaign
How much should I spend?
Life
time
Valu
e ($
)
Allowable
Months as Customer
• Hypothesis Testing Equation• Myth: A vs B, best one wins
• Reality: Look at distribution, segmentation, learn about the market, and dream up C
• Conversion = Desire - Friction
A/B Testing - A Practical Guide
• Behavioral design: Read Cialdini, Dan Ariely, Nir Ayal
• Examples• Headlines
• An authentic solution
• Ask customer what they want
Desire
• UX testing useful for friction• Minimize ambiguity through
good design• Speed
• Call to action - placement, color, shape
• Grouping, contrast, IA
Friction
• Get your market hooked• Their fix = problem solved
• Eliminate all friction
• Test conversion to “fix”• Qualitative methods
Product as Designer Drug