3. Know why, you test and how
you scale what matters:
@MartijnSch
WHAT YOU GET OUT OF THIS?
We’ll be sharing what tools + process we have at The Next Web to get started!
9. But how did we get to
200 A/B Tests a Year?
@MartijnSch#ConversionJam
10. Testing Figures
Data from 2015
2,5CRO Team in 2015
38.000.000Sessions with Tests
0CRO Team in 2014
5-7Tests per week
PricelessAmounts of fun
35%Traffic v.s. Tests
@MartijnSch
11. @MartijnSch
Our roadmap for getting to 200 tests a year
Jan
‘15
Let’s do more testing!
After running 20-30 tests in 2014 we knew we had to scale up.
Resources were low, but with the addition of a web analist we could grow.
Apr
‘15
Hitting some limits
Google Analytics Content Experiments (JS) was our first tool to
use. But soon we hit the limits of the tool.
May
‘15
Setting the vision
After running 40 tests by May we needed a beter process that
could scale and that aligned with our testing vision.
Dec
‘15
Reaching XXX tests
With the testing vision and updated testing program we were able
to run XXX tests in total in 2015.
3 Months
1 Month
7 Months
14. “If you double the number of
experiments you’re going to
double your inventiveness”
- Jeff Bezos
@MartijnSch
15. What makes it different:
Publishers v.s. E-commerce?
@MartijnSch#ConversionJam
16. The opportunity is 350-400 tests a
year. As ROI per winning test is
lower we need to run a lot to have
a positive return.
Efficiency <> Costs
@MartijnSch / #ConversionJam
17. How is our (marketing) team
Structured?
@MartijnSch#ConversionJam
22. We needed to learn,
Knowledge building?
@MartijnSch
You got to love working with people who are like minded.
#ConversionJam
23. Thanks for all your support!
We learn from you!
@MartijnSch / #ConversionJam
24. The more we test the more
we innovate and the higher
inventiveness.
Find the
perfect process
@MartijnSch / #ConversionJam
25. The opportunity is 350-400 tests a
year. As ROI per winning test is
lower we need to run a lot to have
a positive return.
Know your metrics?
@MartijnSch / #ConversionJam
26. @MartijnSch
1
2
Unique Users
Growing our unique user base will
increase the potential to reach Generation T.
2
3 3
Our metrics/KPIs for success;
User Engagement: PV/Session
A higher engagement with our user base will
increase the number of pageviews per session.
User Retention: Returning %
Increasing the percentage or returning users will
increase our brand awareness.
Newsletter Signups
To get a relationship with our users we’d like to
have their email address for the newsletter.
Social Shares
More shares will influence our reach amongst
new and returning users.
27. The opportunity is 350-400 tests a
year. As ROI per winning test is
lower we need to run a lot to have
a positive return.
Do you have a
swimming pool?
@MartijnSch / #ConversionJam
28. The opportunity is 350-400 tests a
year. As ROI per winning test is
lower we need to run a lot to have
a positive return.
Segment Users
@MartijnSch / #ConversionJam
29. Users are the same
@MartijnSch
KNOW YOUR SEGMENTS
X
X
X
30. Users are the same
@MartijnSch
KNOW YOUR SEGMENTS
X
X
X
31. Google Tag Manager (GTM) is
making our lives easier +
scalable.
What tools we use
@MartijnSch / #ConversionJam
32. So we need to improve most of
our communication around
testing.
Communicate more
@MartijnSch / #ConversionJam
33. How did we
overcome our fails?
@MartijnSch
You fail once, you fail twice!
#ConversionJam
34. We didn’t move to any ‘real’
CRO tool but decided to
iterate on our own: GTM <> JS.
GTM
Download:
Github.com/TheNextWeb/CRO
@MartijnSch / #ConversionJam
35. Building our own solutions for
saving testing documentation
+ analysis of tests.
Innovate
@MartijnSch / #ConversionJam
36. We moved most of our
communication towards
Slack. Making the teams more
involved in the whole process.
Always
communicate
@MartijnSch / #ConversionJam
37. We hired two front-end
developers, to make sure in
the long-run we can test
bigger things.
Front-end Dev
@MartijnSch / #ConversionJam
38. 5-7Tests per week
2-4Tests per week
30-40Tests per year
20%More efficient
@MartijnSch
Improvement
What does progress look like?
40. Who’s able to deploy winning
variants within minutes from their
codebase?
Continous
Improvement
@MartijnSch / #ConversionJam
41. This will make segmentation +
automation even easier. In addition
it will decrease the flickering effect.
Back-end
Testing
@MartijnSch / #ConversionJam
42. Testing small doesn’t work, so we
need to grow even bigger. On
average we grow traffic with around
10-20% yearly. At the same rate we
can increase our testing program.
Big > Biggest
@MartijnSch / #ConversionJam
43. TL;DR
Process
1. Have the right company vision
2. Have the right team structure
3. Have access + support from C level
4. Create partnerships with like minded
5. You’ll have trouble as well, nobody’s perfect
6. Don’t be afraid to invent the wheel again
7. Always focus on innovation
@MartijnSch#ConversionJam
47. Sharing brings in more loyalty,
more engagement and
hopefully more traffic.
Share buttons
48. • Variant increase clicks with 16%
• Had a significance + power level of > 99%
• We ran the test for 1x 7+ days
• Targetting over 200.000 users
Winner: Variant
Original:
Variant
51. • Variant increase clicks with 36%
• Had a significance + power level of > 99%
• We ran the test for 7+ days
• Targetting over 300.000 users
Winner: Variant 1
Original: Variant:
53. • Variant increase clicks with 36%
• Had a significance + power level of > 99%
• We ran the test for 2x 14+ days
• Targetting over 500.000 users
Winner: Variant
Original:Variant
54. So what if we remove the
whole sidebar, wouldn’t people
be more focused on the stuff
that is out there?
Big, bigger
55. • Variant decreased clicks with 17% + 14%
• Had a significance + power level of > 99%
• We ran the test for 14+ days
• Targetting over 420.000 users
Winner: Original
56. Now you know how to run
200 A/B Tests a Year?
@MartijnSch
58. “Why didn’t we run 200 tests?”
“What went wrong?”
“Do I need to start crying?”
Boris (CEO)
@MartijnSch / #ConversionJam
59. “Well, if this guy wouldn’t have stuck
his fingers into a lawn mower, YES!”
Missing out on ±20 tests in 2015.
The Lawn Mower
@MartijnSch / #ConversionJam
60. Take Aways:
1. Focus on running X tests a year
2. Improve the process
3. Have the right (free) tools
4. Have the right people + support
5. Document/Communicate everything
6. Know what the future looks like
7. Have fun!
@MartijnSch#ConversionJam