Journal

Speed is a design decision, not a plugin

You cannot cache your way out of a hero video, nine web fonts and a slider. We size the page before we style it, and by the time the caching layer arrives there is very little left for it to do.

SubjectPerformance Published 3 Jun 2026 Reading6 min Written byNadil & Pravin

a waterfall chart taped beside a printout of the page it describes 3/2

Performance 04 / 06

Performance work usually arrives at the end, as a plugin, after the design has already committed to everything expensive. That is the wrong order. By then the only moves left are compression and caching, and both are small next to the decisions made in the first week.

The decisions that matter are unglamorous. How many typefaces, and how many weights of each. Whether the first screen needs a video or a still. Whether the thing that animates has to animate before the page can be read.

We give a page a budget the same way an engineer gives a beam a load: a number, agreed up front, that the design has to live inside. It changes the conversation from "can we make this faster afterwards" to "what does this element cost, and is it worth it".

The result usually looks the same to the client and feels different to the visitor, which is the point. Nobody notices a fast site. They notice a slow one, and they notice it as a feeling about the business rather than as a feeling about the website.

Two founders trained in engineering wrote that last paragraph and had to resist putting a graph under it.

Keep reading Newer What actually belongs above the fold on a trades site Older Stop writing about yourself on the page that sells Index All six pieces Or Have us look at your site