Frontend Performance
A checklist of 48 practices, each with the reasoning behind it.
Performance work has a failure mode: a long checklist, applied uniformly, that makes the code worse and the page no faster. Most items are real, but only a few matter on any given page, and you cannot tell which without measuring first.
So: measure, find the biggest cost, fix that, measure again. The chapters follow the original checklist's priority ordering, which is a reasonable default when you have nothing else to go on.
The short version
Almost every slow page is slow for one of four reasons:
- Too many bytes, usually images or JavaScript.
- Bytes arriving in the wrong order, so the browser waits on something it did not need yet.
- Render-blocking resources, mostly CSS and synchronous scripts.
- A slow server, which no amount of frontend work will fix.
Everything in these chapters is a specific case of one of those.
Chapter 1: High Priority
The 19 items in this chapter account for most of the difference between a fast page and a slow one, and most sites have several outstanding. They fall into three rough clusters:
Targets — page weight, load time and TTFB. Not fixes in themselves, but the numbers the rest of the work moves. Start by measuring them.
Delivery — compression, caching, request count, protocol. Server-side settings, usually a one-time configuration change, usually the cheapest wins available.
Assets — CSS, JavaScript and images. Where the bytes actually are.
The priority tiers come from the original checklist. They are a reasonable default ordering, but your page is not the average page — measure before assuming the high-priority items are where your problem is.
How to read this
Each page states the practice, explains the mechanism behind it, and says when it does not apply. That last part matters: several of the most-repeated performance rules were written for HTTP/1.1 and became actively wrong under HTTP/2.
Where the original and current practice disagree, the page says so — HTTP requests and CSS concatenation are the clearest cases.
Credits
Based on the roadmap.sh Frontend Performance best practices, itself derived from the Front-End Checklist by David Dias. The chapter structure and all 48 items follow the original; the reasoning, code examples and post-HTTP/2 corrections are added.