Engineering
Performance Budgets Are Brand Budgets
There is a conversation that happens on almost every brand site project, usually around week six. The design has gotten richer, the motion more ambitious, the imagery heavier, and someone from our engineering side flags that the page is now crossing our performance budget. In most agencies, this is where a negotiation starts, and the budget loses, because the budget is framed as engineering's preference against the brand's ambition.
We have restructured that conversation entirely, and the reframe is the title of this post. A performance budget is not a technical constraint on the brand. It is part of the brand, as much as the typeface. Sites that respond instantly feel confident, well made, and respectful of the visitor. Sites that lag feel cheap, regardless of how expensive the art direction was. Users cannot see your grid system, but every single one of them can feel 800 milliseconds.
The budget is set in the brand phase
The operational change that makes this real: we set performance budgets during brand strategy, not during development. When we define what a brand should feel like, "fast" gets specified with numbers, alongside the color system, and signed off by the same people.
A typical budget block from a recent engagement:
performance_budget:
lcp: 1.5s # p75, mid-range Android, 4G
inp: 150ms # p75
cls: 0.05
js_critical: 70kb # compressed, initial route
page_weight: 900kb # initial view, hero included
Two details matter here. The numbers are percentile targets on mid-range hardware, because your customers do not browse on your design team's laptops. And the budget lives in the brand guidelines, which changes who defends it. When the heavy hero video shows up in week six, the conversation is no longer engineering versus creative. It is the brand's own standard, agreed by everyone, arbitrating a design decision. The question becomes "how do we achieve this feeling within our budget?" which is a craft question with genuinely interesting answers.
Constraints are a design material
Because here is what the negotiation framing misses: the budget makes the work better, not just faster. Some of the strongest visual moments we have shipped exist because the lazy option was over budget.
- A product reveal built with CSS scroll timelines and three optimized frames instead of a 12 MB background video. It is more distinctive than the video would have been.
- Variable fonts doing the work of six static weights, which pushed the type system toward a tighter, more ownable range.
- An illustration style developed partly because photography at the required quality blew the image budget. It became the most recognized asset in the identity.
Every mature creative discipline works this way. Film has runtimes, print has physical inks and budgets, and nobody calls those compromises. The web pretended for a decade that it had no material constraints, and the result was a slow, interchangeable web. The constraint was always there. Users were paying it.
The business case, briefly
If the craft argument does not move your organization, the commercial one should. The conversion sensitivity to load time is among the most replicated findings in this industry, and it has sharpened as expectations rose. On one retail client's replatform, moving p75 LCP from 3.1 seconds to 1.4 was associated with a measurable lift in conversion that dwarfed the impact of that quarter's creative refresh, and we say that as the people who made the creative refresh. Speed also correlates with crawl efficiency and with how reliably agents and answer engines can extract your content, which is where an increasing share of first impressions now happen.
Fast is a brand value you can implement. It requires only that someone with authority declare it one, write down the numbers, and hold every discipline, including ours, to them. Put it in the guidelines. Right after the logo, ahead of the photography style. That is roughly where users rank it.
Building something this could apply to?
We take on a small number of flagship projects each quarter.
Start a project