Where unseen weight comes from
Manuel Matuzovic's Why 543 KB keep me up at night examines a page whose 543 KB seems disproportionate given its content: one image (a logo of roughly 20 KB) and a single paragraph. As he puts it:
Yes, I know, it depends. 543 KB aren't always bad, but on that specific page there's only a single image (the logo ~20 KB) and a single paragraph. So why then is the page still relatively large, where are the remaining 523 KB coming from?
The answer is JavaScript. The piece also surfaces a less familiar detail: Google's published notion of an ideal DOM, which recommends a page
- contain fewer than 1500 nodes total,
- reach a maximum depth of 32 nodes,
- and avoid any parent node with more than 60 child nodes.
Architecture choices for client-side apps
DebugBear's Performant front-end architecture makes a case for splitting what ships to the browser rather than loading one large bundle:
Bundle splitting will result in more requests being made to load your app. But as long as the requests are made in parallel that's not a big problem, especially if your site served over HTTP/2.
The advice assumes the app is delivered largely as client-side JavaScript. Many sites have abundant low-hanging performance gains available, but the list looks different once an app is client-rendered, and there code-splitting moves near the top of the priorities.
Measurement as an argument
Jeremy Keith, in Telling the story of performance, points out that Web Page Test does more than produce numbers:
Web Page Test is a terrific tool for measuring performance. It can also be used as a story-telling tool.
Because the tool records video of a page loading, that footage can sit side by side with a competitor's and be presented directly to a client.
Performance in revenue terms
CP Clermont's The Impact of Web Performance describes an approach taken at ALDO to quantify the business value of speed without first doing the work of optimising:
In this post, I'll discuss what I did at ALDO to measure the revenue impact of web performance without having to spend time making performance improvements.
The finding — that faster experiences produce more revenue — is expected. The scale is not: users on mobile generated over 3x more revenue, and desktop users nearly 6x more.



