© King.com Ltd 2017 – Commercially confidential
Exploratory Testing for
Everyone
Testit 2017
Page 2
© King.com Ltd 2017 – Commercially confidential
Introduction
Johan Hoberg
• 13+ years of testing
• 10 years at Sony Mobile
• 3+ years at King
• Things that make me happy
• Testing
• Games
• Other stuff (Family, etc.)
Page 3
© King.com Ltd 2017 – Commercially confidential
What is this talk about?
• TL;DR
• We replaced all our scripted testing with
exploratory testing
• The results were very positive in a
number of ways
• I will go into some more detail exactly what
this meant to us and how it affected the team
• At the end I will go through why this is an
important part of a much bigger puzzle
Page 4
© King.com Ltd 2017 – Commercially confidential Page 5
Background
Game
TeamDevelopers, artist, product owner,
Scrum master, QA
Game
Maintain & Develop
Content
Release
New content every 2 weeks
Test
Scripted release regression test
Scripted test cases for new features
© King.com Ltd 2017 – Commercially confidential Page 6
Scripted Release Regression test
3-4 people
doing all testing
200+
Test
cases
”stuck with
testing”
No commitment
No ownership
© King.com Ltd 2017 – Commercially confidential
Scripted Feature Test
• For new features we wrote scripted test
cases, which we then performed once the
feature was integrated
• Some of these tests were then transferred to
the regression test suite
• The actual scripted test cases had little value
except for adding to the regression suite,
since our test specialist ran exploratory
testing anyway, even if creating them and
discussing around the was valuable
Page 7
© King.com Ltd 2017 – Commercially confidential
Why Change?
• We found most release stopping bugs with
the scripted regression tests (although in
general not many new bugs)
• But no one liked it, and it took quite a long
time to finish the tests
• So we wanted to have at least the same
coverage as before, but more fun in less time
• We also wanted to stop creating a lot of new
feature test cases that we then never really
used
Page 8
© King.com Ltd 2017 – Commercially confidential
The Change
• As a test lead I took the decision to try out
using exploratory testing for one release
instead of the old test suite
• We looked at which broader areas the test
cases covered, and created test missions for
those instead
• Everyone in the team participated in testing
together for 1 hour
• We stopped writing test cases for new
features and instead worked on the design
documents with the rest of the team
Page 9
Live OPS
CANVAS
Gameplay
MOBILE
Purchase & Sales
MOBILE
© King.com Ltd 2017 – Commercially confidential
The Result
• Same amount of release stopping bugs
found, but more minor bugs
• Everyone in the team takes more ownership
of quality and testing, not only during the
regression tests, and no one complains
anymore
• What previously took a whole day is done in
one hour because everyone participates and
we tested based on risk, instead of running
“everything”
• No unnecessary test artifacts are created or
maintained for new features, and the design
documents now reflected any risks we saw
Page 10
© King.com Ltd 2017 – Commercially confidential
What it all comes down to
• Motivation
• Risk-based testing
• Not producing waste
• Common ownership of quality and test
• Everyone testing to the extent of their
abilities
Page 11
© King.com Ltd 2017 – Commercially confidential
Bonus Slide: How it all fits together
Page 12
Agile & Scrum
Intrinsic Motivation
Ownership of Quality
High Quality Complex
Products
Mandate & Trust
© King.com Ltd 2017 – Commercially confidential
Bonus Slide: Intrinsic Motivation and Scrum
Self-Determination Theory
• Autonomy
• Competence
• Relatedness
Drive
• Mastery
• Autonomy
• Purpose
Page 13
© King.com Ltd 2017 – Commercially confidential
Bonus Slide: Test Competence and Cynefine
• I believe strongly that
• anyone can perform “obvious” tests
• people with system knowledge can
perform “complicated” tests
• and someone with test competence can
perform “complex” tests
• Complex
• Cause and effect can only be deduced in
retrospect
• probe–sense–respond
Page 14
Thank you!

Testit 2017 - Exploratory Testing for Everyone

  • 2.
    © King.com Ltd2017 – Commercially confidential Exploratory Testing for Everyone Testit 2017 Page 2
  • 3.
    © King.com Ltd2017 – Commercially confidential Introduction Johan Hoberg • 13+ years of testing • 10 years at Sony Mobile • 3+ years at King • Things that make me happy • Testing • Games • Other stuff (Family, etc.) Page 3
  • 4.
    © King.com Ltd2017 – Commercially confidential What is this talk about? • TL;DR • We replaced all our scripted testing with exploratory testing • The results were very positive in a number of ways • I will go into some more detail exactly what this meant to us and how it affected the team • At the end I will go through why this is an important part of a much bigger puzzle Page 4
  • 5.
    © King.com Ltd2017 – Commercially confidential Page 5 Background Game TeamDevelopers, artist, product owner, Scrum master, QA Game Maintain & Develop Content Release New content every 2 weeks Test Scripted release regression test Scripted test cases for new features
  • 6.
    © King.com Ltd2017 – Commercially confidential Page 6 Scripted Release Regression test 3-4 people doing all testing 200+ Test cases ”stuck with testing” No commitment No ownership
  • 7.
    © King.com Ltd2017 – Commercially confidential Scripted Feature Test • For new features we wrote scripted test cases, which we then performed once the feature was integrated • Some of these tests were then transferred to the regression test suite • The actual scripted test cases had little value except for adding to the regression suite, since our test specialist ran exploratory testing anyway, even if creating them and discussing around the was valuable Page 7
  • 8.
    © King.com Ltd2017 – Commercially confidential Why Change? • We found most release stopping bugs with the scripted regression tests (although in general not many new bugs) • But no one liked it, and it took quite a long time to finish the tests • So we wanted to have at least the same coverage as before, but more fun in less time • We also wanted to stop creating a lot of new feature test cases that we then never really used Page 8
  • 9.
    © King.com Ltd2017 – Commercially confidential The Change • As a test lead I took the decision to try out using exploratory testing for one release instead of the old test suite • We looked at which broader areas the test cases covered, and created test missions for those instead • Everyone in the team participated in testing together for 1 hour • We stopped writing test cases for new features and instead worked on the design documents with the rest of the team Page 9 Live OPS CANVAS Gameplay MOBILE Purchase & Sales MOBILE
  • 10.
    © King.com Ltd2017 – Commercially confidential The Result • Same amount of release stopping bugs found, but more minor bugs • Everyone in the team takes more ownership of quality and testing, not only during the regression tests, and no one complains anymore • What previously took a whole day is done in one hour because everyone participates and we tested based on risk, instead of running “everything” • No unnecessary test artifacts are created or maintained for new features, and the design documents now reflected any risks we saw Page 10
  • 11.
    © King.com Ltd2017 – Commercially confidential What it all comes down to • Motivation • Risk-based testing • Not producing waste • Common ownership of quality and test • Everyone testing to the extent of their abilities Page 11
  • 12.
    © King.com Ltd2017 – Commercially confidential Bonus Slide: How it all fits together Page 12 Agile & Scrum Intrinsic Motivation Ownership of Quality High Quality Complex Products Mandate & Trust
  • 13.
    © King.com Ltd2017 – Commercially confidential Bonus Slide: Intrinsic Motivation and Scrum Self-Determination Theory • Autonomy • Competence • Relatedness Drive • Mastery • Autonomy • Purpose Page 13
  • 14.
    © King.com Ltd2017 – Commercially confidential Bonus Slide: Test Competence and Cynefine • I believe strongly that • anyone can perform “obvious” tests • people with system knowledge can perform “complicated” tests • and someone with test competence can perform “complex” tests • Complex • Cause and effect can only be deduced in retrospect • probe–sense–respond Page 14
  • 16.