Mobile A/B testing at LinkedIn: how members shape our apps

Akhilesh Gupta
8 min readbeginner
--
View Original

Overview

The article discusses LinkedIn's approach to mobile A/B testing, highlighting how member feedback shapes app development. It details the use of view-based JSON for dynamic UI rendering and the XLNT platform for targeted experimentation.

What You'll Learn

1

How to implement view-based JSON for mobile app UI rendering

2

Why using a server-side decision process enhances A/B testing accuracy

3

How to leverage the XLNT platform for targeted A/B testing

Prerequisites & Requirements

  • Understanding of A/B testing concepts
  • Familiarity with JSON and mobile app development(optional)

Key Questions Answered

How does LinkedIn use view-based JSON for mobile A/B testing?
LinkedIn employs view-based JSON to dynamically render content in mobile apps without requiring client iterations. This approach allows the server to control UI elements based on predefined views, facilitating easier A/B testing by sending different view types to clients without needing app updates.
What is the XLNT platform and how does it support A/B testing?
XLNT, pronounced 'Excellent', is LinkedIn's A/B testing platform that enables targeted experiments based on user attributes. It allows for real-time adjustments to experiments without client updates, ensuring accurate tracking of user interactions and performance metrics.
What are the benefits of using server-side decision processes in A/B testing?
Using server-side decision processes in A/B testing ensures accurate bucket assignments for users, as the client does not influence the experiment's outcome. This method allows for quick adjustments to experiments and the ability to target users based on contextual information.
How does LinkedIn measure the performance of A/B tests?
LinkedIn measures A/B test performance by sending event data to the XLNT platform, which associates user interactions with relevant metrics like click-through rates and page views. This data is visualized in dashboards to facilitate data-driven decision-making.

Technologies & Tools

Some links below are affiliate links. We may earn a commission if you make a purchase.

A/B Testing Platform
Xlnt
Used for conducting targeted A/B tests based on user attributes.
Backend
Node.js
Formats data for clients based on A/B test results.

Key Actionable Insights

1
Implement view-based JSON to streamline mobile app updates and A/B testing.
This approach allows for dynamic UI changes without requiring app updates, making it easier to experiment with different designs and features.
2
Utilize the XLNT platform for precise targeting in A/B testing.
By leveraging user attributes and contextual information, you can enhance the effectiveness of your experiments and make informed product decisions.
3
Focus on server-side decision-making to improve the accuracy of A/B test results.
This minimizes client-side biases and allows for real-time adjustments to experiments based on user behavior.

Common Pitfalls

1
Relying too heavily on client-side logic for A/B testing decisions.
This can lead to inaccurate results due to biases introduced by the client environment. Instead, server-side decision-making should be prioritized to ensure the integrity of the experiment.

Related Concepts

A/B Testing/Experimentation
Product Design