You have a website. You add some
<script>from a third-party - let's say an analytics tool. Then you add another one, and another one. We scanned the top 5,000 websites in the US, and the average website has 22 different third-party tools implemented in it. We asked - how does this affect the performance of these sites?
There are many ways to measure performance, but we wanted to focus on the metrics that really affect the user experience, and the site SEO. We built a tool based on Lighthouse that analyzes performance with and without the impact of these third-parties. We ran this tool on hundreds of websites and discovered that the average website is 40% slower because of the way these tools are implemented.
We think everyone should care about the performance of their website, so we made this public and now you can test any URL you want.
We are focusing on Time to Interactive - a metric that's becoming an industry standard and tries to answer the question "how long does it take before users can start using my website?". We also included Total Blocking Time and First CPU Idle.
When we say third-party, we don't meant everything that isn't from your domain. If you load React from a CDN, for example, that's an essential part of your website. We are referring to all the things you add to your website after it's finished - marketing pixels, product analysis tools, audience trackers, customer success tools and more. Test it on your website and you'll see exactly what tools we blocked.
Not really. An adblocker could remove ads that are first party - we don't do it on our test. On the other hand, an adblocker would still load that chat bubble third-party tool that you recently added, and we wouldn't. To test your site speed with and without ads, we recommend webtest.app.
We are working this right now and already have some sophisticated tricks to show. It's a closed beta - you can sign up on our homepage.
All default Lighthouse setting, which means mobile with throttling enabled. Each test runs on a AWS machine with 2 CPUs and 4GB RAM. We also keep cache of the results for 24 hours so repeated searches wouldn't hammer the server.
We are cleaning this code up and will soon publish it for you to run and experiment with. Check back soon!
We would love to hear about X and what you think about this tool in general. Feel free to email