"Don't Let Third Parties Slow You Down"
summary
An experiment measures third-party widget cost and delivers a concrete fix: replace the heavy ad script with a tiny loader script that uses document.write into an iframe to fetch and execute the real ad code off the main thread while still placing the ad in the original DOM spot. Live A/B runs across top publishers show median blocking time falls from 47ms to 11ms and 90th-percentile from ~300ms to ~32ms, with several browser-specific gotchas addressed.
AI-written summary · about
chapters (9)
- 0:00Intro
- 0:39Example
- 1:11The Knockout Lab
- 3:38ThirdParty Snippets
- 3:49The Problem
- 4:27The Solution
- 6:24The Framework
- 8:14Script Tag
- 9:23Results
related talks
46mChaperones and curfews: minimising 3rd party impact
41mIt’s My (Third) Party, and I’ll Cry if I Want To
45mHigh Performance Ads - Is It
47mIt’s My (Third) Party, and I’ll Cry if I Want To
53mThird Party Woes
56mThird Party Scripts
54mReducing the Speed Impact of Third-Party Tags
22m#notalljavascript (How to deal with 3rd party JS)