52
Lars Lofgren and Will Kurt Keep Your Gains from A/B Tests Without Killing Them Later May 2014

How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Embed Size (px)

Citation preview

Page 1: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Lars Lofgren and Will Kurt

Keep Your Gains from A/B Tests Without Killing Them Later May 2014

Page 2: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

@larslofgren

Hit me up

Page 3: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 The limits of A/B tests

We’ll cover…

2 The standard solutions

3 Simulations! Woohoo!

#KISSwebinar

4 The 3 strategies of A/B testing that work

5 How we A/B test at KISSmetrics

Page 5: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Limits of A/B Tests

Page 6: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

A/B tests don’t give you perfect decisions.

Page 7: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

No ma!er what you do, you’re never 100% certain

Page 8: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

If we’re not careful, winners aren’t really winners

Page 9: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Your conversions go up… and then they come back down

Page 10: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

The Standard Solution

Page 11: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Run your test until you hit 95% statistical significance.

Page 12: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Go to getdatadriven.com if you need a significance

calculator.

Page 13: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Pick the minimal improvement

Scientific A/B testing:

2 Determine your sample size

3 Determine degree of certainty (95%)

#KISSwebinar

4 Start test but don’t check it early

5 If results aren’t significant, keep control

Page 14: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Martin Goodson’s PDF on poor testing methods:

kiss.ly/bad-testing

Page 15: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

This gives us the best data but not necessarily the

best ROI.

Page 16: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

So how far do we take this?

Page 17: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Simulation Time!

Page 18: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

We modeled several A/B testing strategies.

Using Monte Carlo simulations, we tested different strategies over 1

million observations (people).

Page 19: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Will Kurt gets full credit for all this.

@willkurt

Page 20: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Pick the minimal improvement

The Scientist:

2 Determine your sample size

3 Determine degree of certainty (95%)

#KISSwebinar

4 Start test but don’t check it early

5 If results aren’t significant, keep control

Page 21: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Scientist:

Page 22: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Waits until 80% significance

The Reckless Marketer:

#KISSwebinar

2 Calls a winner as soon as 80% gets hit

Page 23: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Reckless Marketer:

Page 24: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Waits for 95% significance

The Impatient Marketer:

#KISSwebinar

2 Moves on to the next test a#er 500 people

Page 25: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Impatient Marketer:

Page 26: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

The Realist

#KISSwebinar

1 Waits for 99% significance

2 Moves on to the next test a#er 2,000 people

Page 27: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Realist:

Page 28: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

The Persistent Realist

#KISSwebinar

1 Waits for 99% significance

2 Moves on to the next test a#er 20,000 people

Page 29: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Persistent Realist:

Page 30: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

The Blitz Realist

#KISSwebinar

1 Waits for 99% significance

2 Moves on to the next test a#er 200 people

Page 31: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Results for the Blitz Realist:

Page 32: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Let’s compare them using the area under the curve.

Page 33: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

A/B Strategy Scores

Strategy Conditions Score

Scientist Stats like a pro 67759

Reckless Marketer 80% 57649

Impatient Marketer 95% and 500 people 60532

Realist 99% and 2,000 people 67896

Persistent Realist 99% and 20,000 people 68346

Blitz Realist 99% and 200 people 62836

No Testing Testing? NOPE! 50000

Each score is the area under the curve from the simulation. The higher the score, the more

conversions you received.

Page 34: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

0

17500

35000

52500

70000

Persistent Realist Realist Scientist Blitz Realist Impatient Reckless No Testing

50,000

57,64960,532

62,83667,75967,89668,346

A/B Strategy Scores

Page 36: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

3 Strategies

Page 37: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Don’t make decisions at less than 95% significance.

Page 38: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

You’ll waste all the time you spend testing

Page 39: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Be a scientist at 95%

We have 3 viable strategies for making this work:

2 Only make changes at 99%

3 Sloppy 95% but make it up in volume

#KISSwebinar

Page 40: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Pick the minimal improvement

Be a scientist when you have lots of data and resources

2 Determine your sample size

3 Determine degree of certainty (95%)

#KISSwebinar

4 Start test but don’t check it early

5 If results aren’t significant, keep control

Page 41: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

If you don’t have the data or resources to be a

scientist, go fast at 99%.

Page 42: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

And if you still want to play at 95% without being

a scientist, never stop testing.

Page 43: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

How We A/B Test

Page 44: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

First, get volume to 4000+ people/month.

Page 45: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Only make changes at 99% significance.

Page 46: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Let the test run at least 1 week before checking

results.

Page 47: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

If not at 99% a#er two weeks, launch the next

test.

Page 48: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

If the next test isn’t ready, let it keep running while you build the next one.

Page 49: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

The KISSmetrics A/B Testing Strategy

1 Get to 4,000 people/month for test

2 Only change the control if you reach 99%

3 Check results a#er 1 week

4 Launch the next test at 2 weeks

5 Let old tests run if you’re still building

Page 50: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

This strategy isn’t perfect. It’s a balance between good data and speed.

Page 51: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

1 Be a scientist at 95%

Remember the 3 strategies:

2 Only make changes at 99%

3 Sloppy 95% but make it up in volume

#KISSwebinar

Page 52: How to Keep Your Gains from A/B Tests Without Accidentally Killing Them Later

Q&A Time!Lars Lofgren @larslofgren

[email protected]