Optimizely launched our new Stats Engine in January 2015 to power results for all A/B and Multivariate tests run on the platform. Stats Engine lets experimenters see results that are always statistically valid so they can make decisions in real time.
These slides provide an overview of why Optimizely made our new Stats Engine and the problems with traditional statistics that Stats Engine solves. It also provides an introduction to the methods behind Stats Engine, sequential testing and false discovery rate control. Finally, the slides give practical recommendations for how to run an A/B or multivariate test with Stats Engine, and how to decide when you might want to stop your test.
The slides were presented by Darwish Gani, product manager, and Leo Pekelis, statistician, at Optimizely.
See a full recording of the webinar in Optiverse: https://community.optimizely.com/t5/Presentations/Webinar-recording-Stats-Engine-Q-amp-A-webinar-recording-and/ba-p/9776
2. Housekeeping notes
• Chat box is available for questions
• There will be time for Q&A at the end
• We will be recording the webinar for future viewing
• All attendees will receive a copy of slides after the
webinar
4. Objectives
Understand why Optimizely built Stats Engine
Introduce the methods Stats Engine uses to
calculate results
Get practical recommendations for how to test with
Stats Engine
20. Objectives
Understand why Optimizely built Stats Engine
Introduce the methods Stats Engine uses to
calculate results
Get practical recommendations for how to test with
Stats Engine
22. How we did it
• Partnered with Stanford
statisticians
• Talked with customers
• Examined historical
experiments
• Found the best methods
for real-time data
23. What does Stats Engine do?
Provides a principled and mathematical way to calculate your
chance of making an incorrect decision.
24. Sequential Testing False Discovery Rate
• First used in 1940s for military
weapons testing
• Sample size is not fixed in
advance
• Data is evaluated as it’s collected
• First used in 1990s in genetics
• Correct error rates for
multiple goals and variations
• Expected number of false
discoveries
26. Statistical Significance for Digital Experimentation
Continuously Evaluate Test Results
Run many goals and variations
Don’t worry about estimating a
MDE upfront
39. Sample Size + Power Calculations
Focus on creating and running tests
40. Sequential Testing
• Continuously Evaluate Test Results
• Don’t worry about estimating a
MDE upfront
Framework of hypothesis testing that was created to allow the
experimenter to evaluate test results as they come in
41. False discovery rate control
Error rates for a world with many goals and
variations
45. Variations Goal 1 Goal 2 Goal 3 Goal 4 Goal 5
Control
Variation 1
Variation 2
1 False Positive!
Significance Level 90 (False Positive Rate 10%)
1 other variation x goal has a large improvement.
46. Variations Goal 1 Goal 2 Goal 3 Goal 4 Goal 5
Control
Variation 1
Variation 2
1 False Positive!
Significance Level 90 (False Positive Rate 10%)
1 other variation x goal has a large improvement.
1 True Positive!
47. My Report
• Variation 2 is improving on Goal 1
• Variation 1 is improving on Goal 4
“10% of what I
report could be wrong.”
X
50%
• Variation 2 is improving on Goal 1
• Variation 1 is improving on Goal 4
Furthermore, I found the following results.
This leads me to conclude that …
49. • “New York Times has a feature in its Tuesday science section, Take a Number … Today’s
column is in error … This is the old, old error of confusing p(A|B) with p(B|A).”
• Andrew Gelman, Misunderstanding the p-value
• “If I were to randomly select a drug out of the lot of 100, run it through my tests, and
discover a p<0.05 statistically significant benefit, there is only a 62% chance that the
drug is actually effective.”
• Alex Reinhart, The p value and the base rate fallacy
• “In this article I’ll show that badly performed A/B tests can produce winning results
which are more likely to be false than true. At best, this leads to the needless
modification of websites; at worst, to modification which damages profits.”
• Martin Goodson, Most Winning A/B Test Results are Illusory
• “An unguarded use of single-inference procedures results in a greatly increased false
positive (significance) rate”
• Benjamini, Yoav, and Yosef Hochberg. "Controlling the false discovery rate: a practical and powerful approach to
multiple testing." Journal of the Royal Statistical Society. Series B (Methodological) (1995): 289-300. APA
50. • “New York Times has a feature in its Tuesday science section, Take a Number … Today’s
column is in error … This is the old, old error of confusing p(A|B) with p(B|A).”
• Andrew Gelman, Misunderstanding the p-value
• “If I were to randomly select a drug out of the lot of 100, run it through my
tests, and discover a p<0.05 statistically significant benefit, there is only a
62% chance that the drug is actually effective.”
• Alex Reinhart, The p value and the base rate fallacy
• “In this article I’ll show that badly performed A/B tests can produce winning results
which are more likely to be false than true. At best, this leads to the needless
modification of websites; at worst, to modification which damages profits.”
• Martin Goodson, Most Winning A/B Test Results are Illusory
• “An unguarded use of single-inference procedures results in a greatly increased false
positive (significance) rate”
• Benjamini, Yoav, and Yosef Hochberg. "Controlling the false discovery rate: a practical and powerful approach to
multiple testing." Journal of the Royal Statistical Society. Series B (Methodological) (1995): 289-300. APA
51. False Discovery Rate control
Framework for controlling errors that arise from running multiple
experiments at once.
Run many goals & variations
52. What Stats Engine means for you
• You see
fewer, but more accurate conclusive results.
• You can
implement winners as soon as significance is reached.
• You get
• easy experiment workflow.
• reduced unforeseen, and hidden errors.
53. Objectives
Understand why Optimizely built Stats Engine
Introduce the methods Stats Engine uses to
calculate results
Get practical recommendations for how to test with
Stats Engine
55. First, some vocabulary
• Baseline conversion rate
The control group’s expected conversion rate.
• Minimum detectable effect
The smallest conversion rate difference it is
possible to detect in an A/B Test.
• Statistical significance
The likelihood that the observed difference in
conversion rates is not due to chance.
• Minimum sample size
The smallest number of visitors required to
reliably detect a given conversion rate
difference
57. How many visitors do you need to see significant
results?
Visitors needed to reach significance
with Stats Engine
Improvement
5% 10% 25%
Baseline
conversion rate
1% 458,900 101,600 13,000
5% 69,500 15,000 1,800
10% 29,200 6,200 700
25% 8,100 1,700 200
Lower conversion rate, lower effects = more visitors
58. One example of calculating your opportunity cost
12% minimum
detectable effect
61. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
62. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
66. Seasonality
• We DO take into account seasonality while a test is
running.
• We DO NOT take into account future seasonality after an
experiment is stopped.
67. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Difference
Intervals to understand
the types of lifts you
could see.
68. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
72. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Visitors Remaining
to evaluate if waiting
makes sense.
73. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
74.
75. If you’re an organization that can
• Iterate quickly on new variations
• Run lots of experiments
• Have little downside risk of implementing non-winning
variations
then you can likely tolerate a higher error rate.
77. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes • Use Difference
Intervals to measure
risk you take on
80. Should you stop or continue a test?
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Visitors Remaining
to evaluate if waiting
makes sense.
81. Recap
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Difference
Intervals to understand
the types of lifts you
could see.
82. Recap
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Difference
Intervals to understand
the types of lifts you
could see.
Can I afford
to wait?
Continue
Stop
Concede
inconclusive
Yes
No
No
• Use Visitors Remaining
to evaluate if waiting
makes sense.
83. Recap
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Difference
Intervals to understand
the types of lifts you
could see.
Can I afford
to wait?
Continue
Stop
Concede
inconclusive
Yes
No
No
• Use Visitors Remaining
to evaluate if waiting
makes sense.
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
Can I afford
to wait?
• Use Difference
Intervals to measure
risk you take on
84. Recap
Is my test
significant?
Congrats
Can I afford
to wait?
Continue
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
No
Yes
• Use Difference
Intervals to understand
the types of lifts you
could see.
Can I afford
to wait?
Continue
Stop
Concede
inconclusive
Yes
No
No
• Use Visitors Remaining
to evaluate if waiting
makes sense.
Stop
Accept lower
significance
Concede
inconclusive
Yes
No
Can I afford
to wait?
• Use Difference
Intervals to measure
risk you take on
• Use Visitors Remaining
to evaluate if waiting
makes sense.
85. Tuning your testing strategy for your traffic and
business
1%6% 4% 3%
inconclusive inconclusive inconclusiveinconclusive 18%
Looking for small effects?
• Tests take longer to reach
significance
• Might find more winners, if you
are willing to wait long enough
inconclusive
Testing for larger effects?
• Run more tests, faster
• Know when it’s time to move on
to the next idea