By using this site, you agree to the Privacy Policy and Terms of Use.
Accept
World of SoftwareWorld of SoftwareWorld of Software
  • News
  • Software
  • Mobile
  • Computing
  • Gaming
  • Videos
  • More
    • Gadget
    • Web Stories
    • Trending
    • Press Release
Search
  • Privacy
  • Terms
  • Advertise
  • Contact
Copyright © All Rights Reserved. World of Software.
Reading: Why Your A/B Testing Strategy is Broken (and How to Fix It) | HackerNoon
Share
Sign In
Notification Show More
Font ResizerAa
World of SoftwareWorld of Software
Font ResizerAa
  • Software
  • Mobile
  • Computing
  • Gadget
  • Gaming
  • Videos
Search
  • News
  • Software
  • Mobile
  • Computing
  • Gaming
  • Videos
  • More
    • Gadget
    • Web Stories
    • Trending
    • Press Release
Have an existing account? Sign In
Follow US
  • Privacy
  • Terms
  • Advertise
  • Contact
Copyright © All Rights Reserved. World of Software.
World of Software > Computing > Why Your A/B Testing Strategy is Broken (and How to Fix It) | HackerNoon
Computing

Why Your A/B Testing Strategy is Broken (and How to Fix It) | HackerNoon

News Room
Last updated: 2025/08/11 at 9:51 PM
News Room Published 11 August 2025
Share
SHARE

Author:

(1) Daniel Beasley

Table of Links

  1. Introduction

  2. Hypothesis testing

    2.1 Introduction

    2.2 Bayesian statistics

    2.3 Test martingales

    2.4 p-values

    2.5 Optional Stopping and Peeking

    2.6 Combining p-values and Optional Continuation

    2.7 A/B testing

  3. Safe Tests

    3.1 Introduction

    3.2 Classical t-test

    3.3 Safe t-test

    3.4 χ2 -test

    3.5 Safe Proportion Test

  4. Safe Testing Simulations

    4.1 Introduction and 4.2 Python Implementation

    4.3 Comparing the t-test with the Safe t-test

    4.4 Comparing the χ2 -test with the safe proportion test

  5. Mixture sequential probability ratio test

    5.1 Sequential Testing

    5.2 Mixture SPRT

    5.3 mSPRT and the safe t-test

  6. Online Controlled Experiments

    6.1 Safe t-test on OCE datasets

  7. Vinted A/B tests and 7.1 Safe t-test for Vinted A/B tests

    7.2 Safe proportion test for sample ratio mismatch

  8. Conclusion and References

1 Introduction

Randomized controlled trials (RCTs) are the gold standard for inferring causal relationships between treatments and effects. They are widely applied by scientists to deepen understanding of their disciplines. Within the past two decades, they have found applications in digital products as well, under the name A/B test. An A/B test is a simple RCT to compare the effect of a treatment (group B) to a control (group A). The two groups are compared with a statistical test which is used to make a decision about the effect.

Almost all statistical tests for A/B tests rely on fixed-horizon testing. This testing setup involves determining the number of users required for the test, collecting the data, and finally analyzing the results. However, this method of testing does not align with the realtime capabilities of modern data infrastructure and the experimenters’ desires to make decisions quickly. Newly developed statistical methodologies allow experimenters to abandon fixed-horizon testing and analyze test results at any time. This anytime-valid inference (AVI) can lead to more effective use of experimentation resources and more accurate test results.

Safe testing is a novel statistical theory that accomplishes these objectives. As we will see, safe A/B testing allows experimenters to continually monitor results of their experiments without increasing the risk of drawing incorrect conclusions. Furthermore, we will see that it requires less data than standard statistical tests to achieve these results. Large technology companies are currently exploring AVI in limited capacities, but safe testing outperforms available tests in terms of the number of samples required to detect significant effects. This could lead to wide-scale adoption of safe testing for anytime-valid inference of test results.

This thesis contains 6 sections. Section 2 contains an introduction to hypothesis testing, as well as other statistical concepts that are relevant to the reader. It also explains how the inflexibility of classical statistical testing causes issues for practitioners. Section 3 introduces the concepts of safe testing. Furthermore, it derives the test statistics for the safe t-test and the safe proportion test. Section 4 simulates the performance of the safe statistics and compares them to their classical alternatives. Section 5 compares the safe t-test to another popular anytime-valid test, the mixture sequential probability ratio test (mSPRT). Section 6 compares the safe t-test and the mSPRT on a wide range of online experimental data. Finally, Section 7 is devoted to comparing the safe tests to the classical statistical tests at Vinted, a large-scale technology company.

This paper is available on arxiv under ATTRIBUTION-NONCOMMERCIAL-SHAREALIKE 4.0 INTERNATIONAL license.

Sign Up For Daily Newsletter

Be keep up! Get the latest breaking news delivered straight to your inbox.
By signing up, you agree to our Terms of Use and acknowledge the data practices in our Privacy Policy. You may unsubscribe at any time.
Share This Article
Facebook Twitter Email Print
Share
What do you think?
Love0
Sad0
Happy0
Sleepy0
Angry0
Dead0
Wink0
Previous Article Siri's New Features May Include Adding Voice Controls to Apps
Next Article Reddit says its blocking the Internet Archive to stop sneaky AI scrapers accessing its content – News
Leave a comment

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Stay Connected

248.1k Like
69.1k Follow
134k Pin
54.3k Follow

Latest News

MultiBank Group Delivers Record H1 Results With $209M Revenue And MBG Token Driving 7X Returns Since | HackerNoon
Computing
YouTube Has Started Blocking Adblock Plus Users In Chrome – BGR
News
Elon Musk accuses Apple of favoring OpenAI in App Store rankings – 9to5Mac
News
China’s JAC on hiring spree as EV partnership with Huawei revs up: report · TechNode
Computing

You Might also Like

Computing

MultiBank Group Delivers Record H1 Results With $209M Revenue And MBG Token Driving 7X Returns Since | HackerNoon

3 Min Read
Computing

China’s JAC on hiring spree as EV partnership with Huawei revs up: report · TechNode

2 Min Read
Computing

Trust Engines: How Creditcoin Makes Invisible Credit Histories Verifiable | HackerNoon

7 Min Read
Computing

Stellantis’ Chinese partner set to build first European factory in Italy · TechNode

1 Min Read
//

World of Software is your one-stop website for the latest tech news and updates, follow us now to get the news that matters to you.

Quick Link

  • Privacy Policy
  • Terms of use
  • Advertise
  • Contact

Topics

  • Computing
  • Software
  • Press Release
  • Trending

Sign Up for Our Newsletter

Subscribe to our newsletter to get our newest articles instantly!

World of SoftwareWorld of Software
Follow US
Copyright © All Rights Reserved. World of Software.
Welcome Back!

Sign in to your account

Lost your password?