Discover how Skyscanner does Growth across different channels with Growth Director, Michael Briggs.
For more information visit https://medium.com/@Skyscanner/multi-channel-journeys-inside-growth-ddc2ff87af30 and https://www.skyscanner.net/mobile.html for a working example of our product in motion.
2. In Digital Marketing
almost 20 years
Worked in
Edinburgh, Glasgow,
Dubai, London and
Manila
Joined Skyscanner 3
years ago
Director inside
Central Growth
9. At Skyscanner we …
• Believe in growth rather than marketing
• Focus on the problems we solve for users
• Develop systems and mechanisms to deliver solutions to these problems
13. How we measure growth at Skyscanner
User comes to your business from
various channels
14. How we measure growth at Skyscanner
User has a positive experience from
your product/service
15. How we measure growth at Skyscanner
User got value before and returns,
visiting visits multiple times
16. How we measure growth at Skyscanner
User likes experience and product
so much they tell their friends about
you and refers them
17. How we measure growth at Skyscanner
User conducts the activities that
generate revenue
18. How we measure growth at Skyscanner
Channel and Content appropriate for
first time visitors
Channels and Content appropriate
to develop repeated use
24. A multi-channel approach to growth requires :
1. Validate that potential users see this as a problem
2. Understand as deeply as possible the user problem you are solving
3. Develop solution(s) to these problems
4. Create channel visibility in these places
5. Launch responses to re-engage users who don’t get the value
6. Measure everything
7. Restart at stage one
28. Our Hypothesised User Problem Example
“I want to go on holiday to Ibiza but I don’t want to get a good
deal on my flights as I’m tired of dealing with witches and the
dwarves need a break also”
29. Validating the problem
Discovery Channel Platforms Extent of the problem
Search Engines Google, Yandex, Yahoo!,
Naver, Bing
In some markets lots of
questions being asked, in
some markets not as
important
Forums Quora, Yahoo! Answers.
Forums
Number of questions
being asked
Social Facebook, Twitter Ongoing deep
conversations
Ask questions of users User satisfaction, users
testing
Some qualitative data
available
33. Baselines tell us certain things
• The differences in retention between :
• new and returning users
• Mobile and desktop users
• Entry Channel performance
• Natural Multi-channel performance
35. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
36. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
37. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
38. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
39. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
40. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
41. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
42. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
43. IF This Then That Model
QuestionsRetentionActivationAcquisition
Acquired User
through channel
and messaging
Gets Product Value
User Returns
What can we learn
from the cohorts
behavior and
apply?
User Doesn’t
Return
What happened
differently?
Does Something
Else
User Returns
What led the user
to return
User Doesn’t
Return
What does the
behavior tell us we
could have done
better
Bounces User doesn’t return
What could we
improve to deliver
more value?
44. Opportunities
• Email contact to those on newsletter
• Onsite merchandising and personalisation
• Retargetting on Facebook and other platforms
• Potential set of new content required by users
46. Our MVP hadn’t fully solved the problem
“I want to go on holiday to Ibiza but I don’t want to get a good
deal on my flights as I’m tired of dealing with witches and the
dwarves need a break also”
51. Brand Bidding Experiment
No ads showing on brand terms Ads showing on brand terms
Incremental Value
Cannibalisation
Paid Search
Organic Search
52. Two questions we looked to answer
1. What is the impact of brand bidding on brand search?
2. If we pay for sessions is there really a benefit of this activity?
53. Test Design
1. Have a test and control market (with similar characteristics)
2. Phase ad delivery in the test market
3. Change nothing in the control market
4. Analyse results for sessions and revenue
54. How we ran the experiment
Market Phase 1 Phase 2 Phase 3 Phase 4
Test
Control
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
55. How we ran the experiment
Market Phase 1 Phase 2 Phase 3 Phase 4
Test
Control
No Brand
SEM
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
56. How we ran the experiment
Market Phase 1 Phase 2 Phase 3 Phase 4
Test
Control
No Brand
SEM
Brand SEM
Mobile
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
57. How we ran the experiment
Market Phase 1 Phase 2 Phase 3 Phase 4
Test
Control
No Brand
SEM
Brand SEM
Mobile
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
58. How we ran the experiment
Market Phase 1 Phase 2 Phase 3 Phase 4
Test
Control
No Brand
SEM
Brand SEM
Mobile
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
Brand SEM
Mobile +
Desktop
59. Results
1
2
3
Organic Sessions were cannibalised
All of these were recovered through
paid activity
And, sometimes, there was an
incremental gain.
61. Takeaways
1. Put the users needs first when developing any strategy
2. Analytically process results to understand where is most effective
3. Multichannel isn’t always sequential
4. Be willing to be proven wrong!
62. Edinburgh • Glasgow • Singapore • Beijing • Miami • Barcelona • Shenzhen • Sofia •
Budapest
For more example Case Studies about experiments
inside channels and a copy of theseslides, visit:
bit.ly/multichanneljourneys