Core Web Vitals et revenus : le business case de la vitesse

Performance work rarely loses the budget argument because someone disputes the data. It loses because nobody translates milliseconds into money. So here is the translation: page speed is not a technical metric — it is a conversion lever, an SEO input, and a tax on every marketing campaign you run.

This is the business case we present to non-technical stakeholders, with the numbers we use to prioritize what gets fixed.

What the vitals actually measure

Core Web Vitals are Google's three user-experience metrics, and each maps to a business behavior. Largest Contentful Paint (LCP) measures when the main content becomes visible — it decides whether visitors perceive your site as fast or broken. Interaction to Next Paint (INP) measures responsiveness to clicks and taps — it decides whether your site feels sluggish during the moments that matter, like submitting a form.

Cumulative Layout Shift (CLS) measures visual stability — and it is the one most teams underestimate. A button that moves as someone reaches for it does not just annoy users; it produces mis-clicks, accidental purchases, and support tickets.

The conversion math

The published research is consistent in direction even where the exact figures vary: as load time grows from one to three seconds, bounce probability rises sharply; retail conversions measurably drop for every additional second; and Google uses page experience signals in ranking, so slow pages also get fewer visitors before conversion even enters the picture.

The campaign math is what finally lands with executives. If paid traffic lands on a page where one in five visitors bounces before render, you are paying for clicks you never get to show a product to. On meaningful ad spend, fixing a poor LCP is often the highest-ROI marketing action of the quarter — and it is a one-time cost that keeps paying, unlike raising the ad budget.

The fixes that pay for themselves

Not all performance work is equal. We prioritize by return on effort, and the same shortlist applies to almost every site we audit:

• Serve images in modern formats, correctly sized, with explicit dimensions — usually the single largest LCP win
• Render the hero content with priority: preload the critical image and font, defer everything below the fold
• Cut JavaScript shipped on first load — every unused kilobyte is INP debt on mid-range phones
• Fix layout shift by reserving space for ads, embeds, and late-loading banners
• Move to a CDN edge close to your users — physics beats micro-optimization

Make performance a budget, not a project

One-off speed projects decay. A feature shipped six months later quietly re-adds a chat widget and a tracking pixel, and the LCP you fixed is gone. The durable fix is a budget: current vitals plus a defined tolerance, checked in CI on every pull request, with a failing build when a change eats the margin.

Treat the budget like a financial one. Teams can spend it — but they have to notice, and they have to replenish. That single mechanism is the difference between sites that stay fast and sites that get fast every eighteen months, expensively, over and over.