How to set up React A/B testing

Sep 18, 2023

A/B tests help you make your React app better by comparing changes for their impact on key metrics. To show you how to set one up, we will create a basic React app, add PostHog, create an experiment, and implement it to A/B test content in our app.

Creating a React app

First, create our React app using create-react-app and go into the newly created react-ab-test folder.

Terminal
npx create-react-app react-ab-test
cd react-flags

Next, remove the boilerplate code in src/App.js to simplify it to only a title and button we encourage users to press.

JavaScript
// src/App.js
import './App.css';
function App() {
return (
<div className="App">
<h1>We are testing this button:</h1>
<button>Click me!</button>
</div>
);
}
export default App;

Finally, run npm start and go to http://localhost:3000/ to see our new homepage.

React app

Adding PostHog

To use PostHog to manage and track our A/B test, we must install and set it up. We do this by installing the posthog-js library to get access to the React SDK.

Terminal
npm install posthog-js

Once installed, import PostHog into src/index.js and set up a client using our project API key and instance address from our project settings. We can wrap our app in the React PostHogProvider to access PostHog in any component.

JavaScript
import React from 'react';
import ReactDOM from 'react-dom/client';
import './index.css';
import App from './App';
import posthog from 'posthog-js'
import { PostHogProvider } from 'posthog-js/react'
posthog.init('<ph_project_api_key>', {
api_host: 'https://us.i.posthog.com',
})
const root = ReactDOM.createRoot(document.getElementById('root'));
root.render(
<React.StrictMode>
<PostHogProvider client={posthog}>
<App />
</PostHogProvider>
</React.StrictMode>
);

Use the React usePostHog hook in our src/index.js file to capture a custom event when the button is clicked. This will be the goal metric for our A/B test.

JavaScript
// src/App.js
import './App.css';
import { usePostHog } from posthog-js/react
function App() {
const posthog = usePostHog()
const handleClick = () => {
posthog.capture('button clicked');
};
return (
<div className="App">
<h1>We are testing this button:</h1>
<button onClick={handleClick}>Click me!</button>
</div>
);
}
export default App;

Once done, go back to your app, click the button, and you should see a button clicked event captured into your PostHog instance.

Button clicked event

Creating an experiment

With PostHog set up, your React app is ready for the A/B test. To create one, go to the experiments tab in PostHog and click "New experiment." Add a name, feature flag key (like home-button-test), set your goal type to a trend of the button clicked event, and press "Save as draft."

A/B test set up

You can customize it further with a description, secondary metrics, or more variants (for multivariate testing).

Implementing the experiment

With our experiment created, it’s time to add it to our React app. To do this, check the home-button-test flag using the useFeatureFlagVariantKey hook and show our changed content if it returns the value "test."

JavaScript
// src/App.js
import './App.css';
import { useFeatureFlagEnabled, PostHogFeature, usePostHog, useFeatureFlagVariantKey } from 'posthog-js/react';
function App() {
const posthog = usePostHog();
posthog.featureFlags.override('home-button-test', 'test')
const flagValue = useFeatureFlagVariantKey('home-button-test')
const handleClick = () => {
posthog.capture('button clicked');
};
return (
<div className="App">
<h1>We are testing this button:</h1>
{flagValue === 'test' ? (
<button onClick={handleClick}>Sign up for free!</button>
) : (
<button onClick={handleClick}>Click me!</button>
)}
</div>
);
}
export default App;

Alternatively, you can use the React feature component as shown in the How to set up React feature flags tutorial.

To test that it works, override the feature flag with the test value.

JavaScript
//...
posthog.featureFlags.override({'home-button-test': 'test'})
const flagValue = useFeatureFlagVariantKey('home-button-test')
//...

After confirming it works, remove the featureFlags.override call, go back to your experiment in PostHog, and click "Launch." The A/B test is then running to get results on which version is best.

Further reading

Comments