Recipes

Improving Accuracy

Copy for LLMs
Last updated by
Harlan Wilton
 in doc: clean up.

Introduction

Lighthouse performance scores can vary between runs due to network conditions, CPU load, and other factors. These techniques help you achieve more consistent and accurate results.

Multiple Samples Per URL

Run Lighthouse multiple times and average the results for better accuracy:

import { defineUnlighthouseConfig } from 'unlighthouse/config'

export default defineUnlighthouseConfig({
  scanner: {
    samples: 3, // Run 3 scans per URL and average results
  },
})

Reduce Parallel Scans

Limit concurrent workers to reduce CPU contention and improve score consistency:

import { defineUnlighthouseConfig } from 'unlighthouse/config'

export default defineUnlighthouseConfig({
  puppeteerClusterOptions: {
    maxConcurrency: 1, // Single worker for maximum accuracy
  },
})
Combine multiple samples with reduced concurrency for the most accurate results, though this will increase scan time.
Edit this page
Markdown For LLMs
Did this page help you?

UI Customization

Modify Unlighthouse client interface columns and display to show custom metrics and data.

Large Sites

Strategies and configuration for efficiently scanning large websites with thousands of pages.