Third Party Woes

on YouTube ↗53m

" summary

Audit shows one checkout widget injects ~800 KB of JS and removing it cuts total blocking time by ~2.4s, a chatbot removal improves LCP by ~2.24s, and moving scripts into a worker with PartyTown in a worker initially slashes main‑thread contention but breaks when GTM injects scripts so it gets removed. Tools used: requestmap/webpagetest/lighthouse and the Core Web Vitals technical report. Emerging fixes: server/edge rewrites and measured total blocking time comparisons.

AI-written summary · about

" chapters (15)

1. 0:00Introduction
2. 1:26Early web performance insights
3. 3:12Third-party scripts challenges
4. 8:09Analyzing third-party performance
5. 17:24Finding problematic scripts
6. 19:11Evaluating necessity and alternatives
7. 22:22Unavoidable dependencies frustration
8. 25:57Website owners' struggles analogy
9. 27:19Party Town and Cloudflare Workers
10. 31:14Other performance strategies
11. 34:02Third-party scripts data lack
12. 37:59Core Web Vitals technical report
13. 39:02Call to action for community
14. 44:45Third-party-free future dream
15. 45:15Q&A

" related talks

• 09mOffloading Main Thread: Improving Web Performance with PartyTownAtiksh Srivastava · React Nexus 2023
• 46mChaperones and curfews: minimising 3rd party impactRyan Townsend · #PerfMatters Conference 2019
• 54mReducing the Speed Impact of Third-Party TagsAndy Davies · performance.now() 2022
• 31mIt’s My (Third) Party and I'll Cry If I Want ToHarry Roberts · DeltaV Conference 2018
• 13m"Don't Let Third Parties Slow You Down"Michael Kleber, Arvind Jain · Velocity 2010
• 52mIt’s My (Third) Party, and I’ll Cry if I Want ToHarry Roberts · Cold Front 2018
• 50mIt’s My (Third) Party, and I’ll Cry if I Want ToHarry Roberts · JS Fest 2019 Spring
• 41mIt’s My (Third) Party, and I’ll Cry if I Want ToHarry Roberts · JSHeroes 2019