Building Startups & Minimum Viable
Products

@Ben_Hall
Ben@BenHall.me.uk
Cornershop.io
Hacker in Residence and Partner at Cornershop / #1seed
You talk about it, we ship it.
#craftsmanship

@Ben_Hall
Ben@BenHall.me.uk
Cornershop.io
Hacker in Residence and Partner at Cornershop / #1seed
Let’s make some money
#craftsmanshipisdead

@Ben_Hall
Ben@BenHall.me.uk
Cornershop.io
Hacker in Residence and Partner at Cornershop / #1seed
Who am I?
• Hacker in Residence at Cornershop / #1seed
– Meerkatalyst / MaydayHQ (Co-founder)
– Swapit (Lead Dev / CTO)
– 7digital
– Red Gate Software

• Multiple open source and side projects
• @Ben_Hall or Ben@BenHall.me.uk
Agenda

Idea

Build

Release
How I see the world
With a bit of a rant and various
lessons learned along the way
Startup Mindset
Validate / Invalidate a concept as
fast as possible
“Lean”
Build

Learn

Measure
Not about being cheap
Lessons learned
It’s good to fail, as long as you
fail fast
But not so fast that it’s
impossible to succeed
What is a startup?
A company that is confused about
1) what its product is
2) who its customers are
3) how to make money
Validating The Idea
What happens if you don’t have
an idea?
Find a space, theme, problem that
you personally find interesting
Betaworks, Science Inc, Rocket
Internet
Without a core desire, you’ve already lost the game

BELIEVE IN THE VISION
People don't buy what you do;
they buy why you do it.

http://www.flickr.com/photos/atomicpuppy/2132073976/
Easier to go from shit to great
than nothing to great.
Going to Church (a punt) vs. Science (running a process)

CORNERSHOP PROCESS
Leave the building
Understand potential customer
acquisition approaches early
Identify a Minimum Viable Customer
Segment
•
•
•
•

Influencers
Users of competitor products
Potential new users
People in different verticals with similar business
models
• Understand industry, customer segments
• Test different value props, identify which
connects best
Always discuss previous experiences.
People are rubbish at predicting the
future
LISTEN TO USERS/CUSTOMERS NOT
ADVISORS / VC. CUSTOMERS PAY.
Iterate, update, log insights
Go back to BMC and update.
Keep previous copies for history.
People seem interested? You’ll
need a name & brand
Naming is the hardest problem
Can kill days and suck motivation
Google

Airbnb

Spotify

YouPorn

Apple
Ryanair
KEEP TAGLINES SIMPLE
BUT SAY WHAT YOU DO!

DON’T SAY YOU’RE DISRUPTION
OR INNOVATING
LAME!!! Only others can say that
http://insideintercom.io/what-everyoneneeds-to-know-about-disruption/
CustDev can only get you so far
At this point you should have a
concept / vision clear in your own
head
The Build
What is a MVP?
• “An MVP is an experiment that tests a critical,
falsifiable hypothesis of your business” Devin
Hunt

https://speakerdeck.com/devinhunt/mvps-in-practice
Fake it until you make it
Lesson learned from
Rate it Slate it
Prototyped Functionality
Took 2-3 hours to go from a concept to
learning valuable insights
Avoid writing code if you can
• Email / blog first startups are cool!!
• Sunrise (Just raised $2.2m, started as morning
email of your day’s schedule each day)
• Mattermark
Speed of delivery is key
•
•
•
•

Beg, steal, borrow – just get it done!
Ability to learn
Should be based around your vision
Lesson from “Project X”: Took too long to
release because the aim wasn’t to learn about
customers but make money. Missed a
number of (commercial) opportunities.
Failing? Likely.
Don’t reinvent the wheel

TOO MANY FUN THINGS IN LIFE TO
WRITE BORING CODE
Do you need a full application?
WebFlow.com / strikingly.com
Fake it?
• Proto.io
Really need to build…
• Foundation over bootstrap etc
• KISS!! Do you really need EmberJS, Backbone
etc etc etc?
Build on the shoulders of giants
Community
NodeJS  *AMAZING*
ElasticSearch  *AMAZING*
4SQ API, Screen scraping, hidden APIs –
whatever is required to get the job done.

KEEP IT SIMPLE. KEEP IT DIRTY.
Low traffic, dyno is paused. First user needs to spin it up. Keep alive script

A WORD OF WARNING ABOUT
HEROKU
Paul Stack rocked my world – Vagrant
+ Rackspace Cloud / AWS EC2
Cult of the Software Craftsman
• Code Quality is not a feature!
• Do you really need 80% test coverage? What
value is that actually adding?
• Do you really need that abstraction? That IoC?
That level of separation? That ability to scale?
• Is that really going to change your world?
BULLSH*T
WHO CARES ABOUT CLEAN CODE IF
THERE ARE NO SALES!
IF DEVELOPERS WERE IMPACTED BY
REVENUE THEN THEY WOULDN’T
SPEND 10 DAYS WRITING CLEAN CODE
Write tests / code / shell scripts to speed up feedback loop – not about
long-term.

STOPPED DOING TDD A LONG TIME
AGO
DON’T BE AFFAIRD TO THROW
CODE AWAY
Building a startup?
Don’t turn into a developer!
• This isn’t an exercise in learning new
technologies.
• It’s an exercise in building businesses
• Don’t confuse the two.
Lesson learned from
Meerkatalyst
Built an over complex system when a shell
script would have done
Lesson learned from Mayday
Stopped the MVP, started scaling (technical
backend), didn’t have product/market fit.

Massive Fail
Do things that move the needle
What are you trying to learn? What
value are you adding?
The
Release
Job done! If you build it they will come

Must be true, Hollywood said so
Customers
Metrics
Application

TIME TO LISTEN
Metrics
• Collect everything
• Visualise key data
• Have key metrics that you want to track
Librato

Librato - Pretty
Track User Events / Actions
Your own database is most valuable source. Use it!
Metrics became the tests
• Ensure the system is
working as expected
• Alerts when the system
stopped working outside
of normal bounds
Pirate Metrics - AARRR
Drive Traffic
•
•
•
•

PR (Hacker News, Techcrunch, The Next Web)
Buy traffic (PPC)
Social Media
Email still king

• Piggyback off others
– Paypal > Ebay
– Airbnb > Craigslist
Speak to people using the product
• Red Gate UX team
• Watch, Listen, Learn
• Introduce explicit touch points in the
application for reaching out
– Rate it Slate it inbox beta list
• Do people want the feature?
• Can we build a email list of people who are actively
engaging with the product
Quickly qualify a lead & call
A/B Tests?
• Waste of time at the early stage.
• Complex to configure, not enough traffic to
make them statistically significant.
– Mayday A/B tests
• Took ages to get data, could have just asked people
Finally…
Start with something, iterate.
Have a vision, a goal, a desire and
execute on it.
Value Driven Development
Do things that people care about!
Get feedback. Not working?
Change it. Kill it. Move on.

Building startups people care about is
amazing!
Working on ones no-one cares about is
sole destroying.
@Ben_Hall
Ben@BenHall.me.uk
Blog.BenHall.me.uk
Cornershop.io

DevDay 2013 - Building Startups and Minimum Viable Products

  • 1.
    Building Startups &Minimum Viable Products @Ben_Hall Ben@BenHall.me.uk Cornershop.io Hacker in Residence and Partner at Cornershop / #1seed
  • 2.
    You talk aboutit, we ship it. #craftsmanship @Ben_Hall Ben@BenHall.me.uk Cornershop.io Hacker in Residence and Partner at Cornershop / #1seed
  • 3.
    Let’s make somemoney #craftsmanshipisdead @Ben_Hall Ben@BenHall.me.uk Cornershop.io Hacker in Residence and Partner at Cornershop / #1seed
  • 4.
    Who am I? •Hacker in Residence at Cornershop / #1seed – Meerkatalyst / MaydayHQ (Co-founder) – Swapit (Lead Dev / CTO) – 7digital – Red Gate Software • Multiple open source and side projects • @Ben_Hall or Ben@BenHall.me.uk
  • 5.
  • 6.
    How I seethe world With a bit of a rant and various lessons learned along the way
  • 7.
  • 8.
    Validate / Invalidatea concept as fast as possible
  • 9.
  • 10.
    Not about beingcheap Lessons learned
  • 11.
    It’s good tofail, as long as you fail fast
  • 12.
    But not sofast that it’s impossible to succeed
  • 13.
    What is astartup? A company that is confused about 1) what its product is 2) who its customers are 3) how to make money
  • 14.
  • 15.
    What happens ifyou don’t have an idea? Find a space, theme, problem that you personally find interesting
  • 17.
    Betaworks, Science Inc,Rocket Internet
  • 18.
    Without a coredesire, you’ve already lost the game BELIEVE IN THE VISION
  • 19.
    People don't buywhat you do; they buy why you do it. http://www.flickr.com/photos/atomicpuppy/2132073976/
  • 20.
    Easier to gofrom shit to great than nothing to great.
  • 21.
    Going to Church(a punt) vs. Science (running a process) CORNERSHOP PROCESS
  • 24.
    Leave the building Understandpotential customer acquisition approaches early
  • 25.
    Identify a MinimumViable Customer Segment • • • • Influencers Users of competitor products Potential new users People in different verticals with similar business models • Understand industry, customer segments • Test different value props, identify which connects best
  • 26.
    Always discuss previousexperiences. People are rubbish at predicting the future
  • 27.
    LISTEN TO USERS/CUSTOMERSNOT ADVISORS / VC. CUSTOMERS PAY.
  • 28.
    Iterate, update, loginsights Go back to BMC and update. Keep previous copies for history.
  • 29.
    People seem interested?You’ll need a name & brand Naming is the hardest problem Can kill days and suck motivation
  • 30.
  • 31.
    KEEP TAGLINES SIMPLE BUTSAY WHAT YOU DO! DON’T SAY YOU’RE DISRUPTION OR INNOVATING LAME!!! Only others can say that http://insideintercom.io/what-everyoneneeds-to-know-about-disruption/
  • 32.
    CustDev can onlyget you so far At this point you should have a concept / vision clear in your own head
  • 33.
  • 34.
    What is aMVP? • “An MVP is an experiment that tests a critical, falsifiable hypothesis of your business” Devin Hunt https://speakerdeck.com/devinhunt/mvps-in-practice
  • 35.
    Fake it untilyou make it
  • 36.
    Lesson learned from Rateit Slate it Prototyped Functionality Took 2-3 hours to go from a concept to learning valuable insights
  • 37.
    Avoid writing codeif you can • Email / blog first startups are cool!! • Sunrise (Just raised $2.2m, started as morning email of your day’s schedule each day) • Mattermark
  • 38.
    Speed of deliveryis key • • • • Beg, steal, borrow – just get it done! Ability to learn Should be based around your vision Lesson from “Project X”: Took too long to release because the aim wasn’t to learn about customers but make money. Missed a number of (commercial) opportunities. Failing? Likely.
  • 39.
    Don’t reinvent thewheel TOO MANY FUN THINGS IN LIFE TO WRITE BORING CODE
  • 40.
    Do you needa full application? WebFlow.com / strikingly.com
  • 41.
  • 42.
    Really need tobuild… • Foundation over bootstrap etc • KISS!! Do you really need EmberJS, Backbone etc etc etc?
  • 43.
    Build on theshoulders of giants Community NodeJS  *AMAZING* ElasticSearch  *AMAZING* 4SQ API, Screen scraping, hidden APIs – whatever is required to get the job done. KEEP IT SIMPLE. KEEP IT DIRTY.
  • 44.
    Low traffic, dynois paused. First user needs to spin it up. Keep alive script A WORD OF WARNING ABOUT HEROKU
  • 45.
    Paul Stack rockedmy world – Vagrant + Rackspace Cloud / AWS EC2
  • 46.
    Cult of theSoftware Craftsman • Code Quality is not a feature! • Do you really need 80% test coverage? What value is that actually adding? • Do you really need that abstraction? That IoC? That level of separation? That ability to scale? • Is that really going to change your world?
  • 47.
  • 48.
    WHO CARES ABOUTCLEAN CODE IF THERE ARE NO SALES!
  • 49.
    IF DEVELOPERS WEREIMPACTED BY REVENUE THEN THEY WOULDN’T SPEND 10 DAYS WRITING CLEAN CODE
  • 50.
    Write tests /code / shell scripts to speed up feedback loop – not about long-term. STOPPED DOING TDD A LONG TIME AGO
  • 51.
    DON’T BE AFFAIRDTO THROW CODE AWAY
  • 52.
    Building a startup? Don’tturn into a developer! • This isn’t an exercise in learning new technologies. • It’s an exercise in building businesses • Don’t confuse the two.
  • 53.
    Lesson learned from Meerkatalyst Builtan over complex system when a shell script would have done
  • 54.
    Lesson learned fromMayday Stopped the MVP, started scaling (technical backend), didn’t have product/market fit. Massive Fail
  • 55.
    Do things thatmove the needle What are you trying to learn? What value are you adding?
  • 56.
  • 57.
    Job done! Ifyou build it they will come Must be true, Hollywood said so
  • 58.
  • 59.
    Metrics • Collect everything •Visualise key data • Have key metrics that you want to track
  • 60.
  • 61.
    Track User Events/ Actions Your own database is most valuable source. Use it!
  • 62.
    Metrics became thetests • Ensure the system is working as expected • Alerts when the system stopped working outside of normal bounds
  • 63.
  • 64.
    Drive Traffic • • • • PR (HackerNews, Techcrunch, The Next Web) Buy traffic (PPC) Social Media Email still king • Piggyback off others – Paypal > Ebay – Airbnb > Craigslist
  • 65.
    Speak to peopleusing the product • Red Gate UX team • Watch, Listen, Learn • Introduce explicit touch points in the application for reaching out – Rate it Slate it inbox beta list • Do people want the feature? • Can we build a email list of people who are actively engaging with the product
  • 66.
    Quickly qualify alead & call
  • 67.
    A/B Tests? • Wasteof time at the early stage. • Complex to configure, not enough traffic to make them statistically significant. – Mayday A/B tests • Took ages to get data, could have just asked people
  • 68.
  • 69.
    Start with something,iterate. Have a vision, a goal, a desire and execute on it.
  • 70.
    Value Driven Development Dothings that people care about!
  • 71.
    Get feedback. Notworking? Change it. Kill it. Move on. Building startups people care about is amazing! Working on ones no-one cares about is sole destroying.
  • 72.

Editor's Notes

  • #5 I have fun with startups
  • #6 http://www.flickr.com/photos/newyorkbaltimore/8010607852/sizes/k/in/photolist-dcSsCu/Taking everything you’ve heard this week and applying it to real businesses with the aim of making *money*
  • #8 Stages of a startupBuild, measure, leanDon’t be scared to failDon’t be scared to kill it if it’s not workingBe public, be visible, talk to everyoneDon’t fail too early!
  • #17 Meerkatalyst example Thought I understood customer problem. Problem I had personally while at 7digital, knew others had it, ran with it as a side project before joining Springboard startup accelerator
  • #19 Ideas by themselves are worthless100% on executing the visionPrevious company tried to split attention across 4 company streams. It doesn’t work.Without a core desire, you’ve already lost the game
  • #22 Ideas by themselves are worthless100% on executing the visionPrevious company tried to split attention across 4 company streams. It doesn’t work.Without a core desire, you’ve already lost the game