How many times a developer has
to implement a feature that will
be the next game changer?
Slide 3
Slide 3 text
Do you
remember
Google Wave?
Slide 4
Slide 4 text
Do you
remember
Google Wave?
Slide 5
Slide 5 text
No content
Slide 6
Slide 6 text
Or Apple’s Ping?
Slide 7
Slide 7 text
Or Apple’s Ping?
Slide 8
Slide 8 text
No content
Slide 9
Slide 9 text
If the biggest companies in
the world sometimes fails,
why wouldn’t we fail?
“
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
No content
Slide 12
Slide 12 text
Agenda ● Build
● Measure
● Learn
Slide 13
Slide 13 text
Build
Slide 14
Slide 14 text
From developer
point of view
● Move fast
● Unit tests
● Automatic tests
● Technical debts tasks
● Release
Slide 15
Slide 15 text
● Also called MVP
Minimum Viable Product
Slide 16
Slide 16 text
● Also called MVP
● Only includes core part of the feature and no
more
Minimum Viable Product
Slide 17
Slide 17 text
● Also called MVP
● Only includes core part of the feature and no
more
● Allows to verify correct or wrong assumptions
Minimum Viable Product
Slide 18
Slide 18 text
● Also called MVP
● Only includes core part of the feature and no
more
● Allows to verify correct or wrong assumptions
● Accelerate learning
Minimum Viable Product
Slide 19
Slide 19 text
● Also called MVP
● Only includes core part of the feature and no
more
● Allows to verify correct or wrong assumptions
● Accelerate learning
● Is often less expensive to implement an MVP
Minimum Viable Product
Slide 20
Slide 20 text
Is the smallest thing you can build that delivers
customer value (and as a bonus captures some of
that value back)
Minimum Viable Product
Slide 21
Slide 21 text
No content
Slide 22
Slide 22 text
Begins the learning process,
rather than ending it
“
Slide 23
Slide 23 text
When in doubt: simplify
“
Slide 24
Slide 24 text
Use case:
Lookalikes
Slide 25
Slide 25 text
Measure
Slide 26
Slide 26 text
Of course, you can’t measure if
you don’t have a tool or SDK to
report statistics
In-House
Solution
Slide 27
Slide 27 text
Some techniques that will help
you to measure and compare
features in your application
Slide 28
Slide 28 text
Split tests
● Also known as A/B/C/… tests
Slide 29
Slide 29 text
Split tests
● Also known as A/B/C/… tests
● Different versions of the feature with identical
purpose introducing variants between them
Slide 30
Slide 30 text
Split tests
● Also known as A/B/C/… tests
● Different versions of the feature with identical
purpose introducing variants between them
● Control the amount of users on every version
Slide 31
Slide 31 text
Split tests
● Also known as A/B/C/… tests
● Different versions of the feature with identical
purpose introducing variants between them
● Control the amount of users on every version
● Test different hypothesis to detect the most
effective
Slide 32
Slide 32 text
Use case: Own profile
Slide 33
Slide 33 text
Use case: Paywall
Slide 34
Slide 34 text
Remote feature
flag/toggles
Provides the ability to remotely control
(enable or disable) features that are in
production.
Slide 35
Slide 35 text
Learn
Slide 36
Slide 36 text
Is time to think if you want to
discard or persevere
Slide 37
Slide 37 text
Discard the feature
Is possible that the feature didn’t
work well, in that case is time to think
in removing it because is could be
distracting the user
Slide 38
Slide 38 text
Tune the feature
Use all the measurements, validate
the results and tune the feature.
Add possible modifications,
improvements, polish UI, ...
Slide 39
Slide 39 text
Use case:
Lookalikes
Slide 40
Slide 40 text
Use case:
Lookalikes
Slide 41
Slide 41 text
Gain visibility
Add more entry points
placements
Slide 42
Slide 42 text
Use case:
Female
walkthrough
Initial subtle entrypoint
Slide 43
Slide 43 text
Use case: Female walkthrough
Slide 44
Slide 44 text
Some extra tasks should be done
before the next iteration
Slide 45
Slide 45 text
Don’t forget the
technical debt
● Maybe some technical debt was
created during the
implementation of the MVP
Slide 46
Slide 46 text
Don’t forget the
technical debt
● Maybe some technical debt was
created during the
implementation of the MVP
● Do some extra clean up if
possible
Slide 47
Slide 47 text
Don’t forget the
technical debt
● Maybe some technical debt was
created during the
implementation of the MVP
● Do some extra clean up if
possible
● Split tests non-winner variants
code to be removed
Slide 48
Slide 48 text
Summarize ● Build
■ MVP, start learning process
● Measure
■ Split tests, remote feature flags
● Learn
■ Discard or persevere
■ Tech debt and clean up pending
tasks