Reports from Google's Lighthouse tool report web performance metrics based on a specific network speed. Lighthouse uses a particular network throttling method called simulated throttling.
This article explains what Lighthouse simulated throttling is and how it can lead to inaccurate site speed metrics. We'll also look at alternative ways to test your site speed.
What is Lighthouse and what tools are built on top of it?
Lighthouse is a free tool developed by Google that powers many other services under the hood:
- PageSpeed Insights (Lab Data)
- The Chrome DevTools Lighthouse tab
- Commercial tools like DebugBear, GTmetrix, or Calibre

Why do page speed tools throttle the network connection?
Web performance tests are often run on a computer with a fast network connection, for example in a data center or on a wifi connection. Page speed testing tools slow down the network in order to better show how a real user might experience a website, for example if a user is on a slow mobile connection.
Network throttling also ensures that metrics are more consistent, as the same network speed is used to run every test.
What is simulated throttling?
There are several different ways to slow down the network. Simulated throttling is one of them, and it's what Lighthouse uses by default.
With simulated throttling the initial site speed data is collected on a fast connection. Based on this data Lighthouse then estimates how quickly the page would have loaded on a different connection.
For example, if a page takes 2 seconds to render on a fast connection, Lighthouse might report a value of 6 seconds on a mobile device.
Simulated throttling provides low variability and makes tests quick and cheap to run. However, it can also lead to inaccuracies as Lighthouse doesn't fully replicate all browser features and network behaviors.
Check out this article for an in-depth look at how simulated throttling works in Lighthouse.
What tools use simulated throttling?
Simulated throttling is the default for Lighthouse, but Lighthouse also supports other throttling methods.
The lab diagnostics on PageSpeed Insights always use simulated throttling. The Chrome DevTools Lighthouse tab and the Lighthouse command line interface (CLI) use simulated throttling by default but also provide alternative options.
Dedicated commercial performance testing tools like DebugBear or WebPageTest solutions generally don't use simulated throttling.
PageSpeed Insights provides both lab data collected using Lighthouse and real-user data from the Chrome User Experience Report. Real-user data does not use any type of artificial throttling.
How can I tell what throttling method is used to run a test?
The bottom of the full Lighthouse report shows the test settings used to test the page. Hover over the network throttling details to see what type of throttling was used.

What are observed metrics?
Observed metrics are real measurements that were collected by Chrome. When simulated throttling is not used these values are equal to the final values reported by Lighthouse.
When simulated throttling is used then the reported values are generally worse than the observed values, as a slower connection is simulated. If the reported values are better than the observed values this usually indicates an inaccuracy in the throttling simulation.
You can find the observed value in the full Lighthouse report JSON. In this example:
- Lighthouse measured a First Contentful Paint score of 695 milliseconds.
- After running the simulation, the reported value was 1,457 milliseconds.

For PageSpeed Insights, our Site Speed Chrome extension surfaces the observed metrics in the UI.

