Core Web Vitals and the JavaScript budget

Core Web Vitals put numbers to arguments we had been having for years about what actually matters on web pages. Largest Contentful Paint. Cumulative Layout Shift. Measured in the field on real devices, not the lab on fast machines.

Three projects that looked fine to us in testing suddenly had problems when we looked at the field data from real users. Real users on real connections were seeing delayed paints and layout shifts. Lab tests had hidden the variability and user experience impact.

We traced the slow paths through the waterfall and found the culprit was rarely what we expected. It was not the large hero image or the framework bundle size. It was third-party scripts – tags nobody remembered adding and nobody owned responsibility.

Those tags existed because of business requirements and stakeholder demands. Analytics vendors, heat mapping tools, ads. Each one was there for a stated reason. But the cost was invisible until Google put a number on it and ranked it.

We did not remove all of them. We moved them around, deferred them, and added a kilobyte budget per page checked on every pull request. It became a constraint like any other metric and deployment gate.

Audits once a quarter never worked to enforce standards. A constraint that fails a deployment if violated actually works. Priority shifted from yearly reviews to something that mattered every day.