The Stonemason's Unneeded Quarry: On the Hollow Weight of an Aggregated Score

There’s a single number that haunts the dreams of many a developer, a project manager, and a C-suite executive. It’s a number that has been elevated to the status of gospel, a north star by which entire digital strategies are steered. I’m speaking, of course, of the Core Web Vitals score, that aggregate judgment rendered by tools like PageSpeed Insights or Lighthouse. We chase after it with the desperate fervor of a prospector hunting gold, convinced that if we can just push it from ‘Needs Improvement’ to ‘Good’, we will have achieved web performance nirvana. But I fear we are becoming stonemasons who, tasked with building a graceful arch, become obsessed with the sheer tonnage of stone in the quarry rather than the precise placement of each keystone.

The problem isn’t with the metrics themselves. Largest Contentful Paint (LCP), Cumulative Layout Shift (CLS), and Interaction to Next Paint (INP) are each, in their own right, invaluable lenses through which to view user experience. They point to real, tangible frustrations: a screen that paints too slowly, text that jumps as you’re about to click it, a button that feels sluggish and unresponsive. These are the genuine problems. The aggregated score, however, is an abstraction. It’s a mathematical reduction of complexity into a single, seemingly objective grade. And in doing so, it creates a dangerous illusion of simplicity.

The False Economy of the Composite

This illusion leads us astray in two significant ways. First, it encourages a kind of performance myopia. A team, seeing a ‘poor’ score, might embark on a frantic optimization spree, shaving milliseconds off a script here, preloading a resource there. They might succeed in moving the needle, celebrating as the score turns green. But did they solve the problem that actually matters to their users? Perhaps not. They may have optimized a landing page for a metric while ignoring a truly dreadful customer journey on a subsequent, more complex application page that users actually complain about. The high score on the homepage becomes a vanity metric, a trophy that distracts from the real, muddy work of improving the entire experience.

Second, and more subtly, the pursuit of a perfect score can lead to perverse optimization. It can incentivize architecture choices that are technically clever but experientially hollow. For instance, a developer might artificially deflate their LCP by switching a meaningful hero image to a bland, single-colour block, or cripple their site’s interactivity by serving a static, non-interactive shell to ace the initial paint metrics. The numbers look brilliant. The user, however, encounters a sterile, jarring facade. We’ve built the arch with perfectly measured, uniformly shaped stones, but forgotten the opening it was meant to frame.

True performance craftsmanship isn’t about acing a test. It’s about empathy. It’s about understanding the rhythm and flow of a real person’s interaction with your work. It requires looking beyond the hammer of a single score and picking up the full set of chisels: real user monitoring (RUM) to see actual frustrations, focused sessions with real devices on real networks, and a deep, qualitative understanding of what your users are trying to accomplish. The individual Core Web Vitals are excellent chisels in that set. But the aggregate score is just the weight of the toolbox. A heavy toolbox doesn’t mean you’re a better craftsman; it only means you’re carrying more. The real art lies in knowing which tool to use, and when, to shape an experience that feels not just fast, but fluid, stable, and—most importantly—human.

Notes & further reading

A few pages I came back to while writing this: