A South Florida store owner messaged me a few months ago. Traffic was fine and ad spend was steady, but sales had flatlined. On a real phone over cellular, the product image took almost six seconds to appear, and "Add to Cart" jumped down the screen just as people tapped it. Roughly two in three mobile visitors left before the page was usable.
Slow sites don't throw errors. They just leak money. Core Web Vitals are Google's numbers for that experience, and in 2026 they map more directly to revenue than almost anything else you control. Here's what they are, how much they matter, and what I'd fix first.
What Core Web Vitals actually measure in 2026
Each metric matches a moment of frustration you've felt yourself: waiting for the page, tapping and getting nothing, or hitting the wrong button because the layout moved. INP replaced FID in March 2024 and is much harder to game, because it measures essentially every interaction, not just the first.
| Feature | Good | Needs improvement | Poor |
|---|---|---|---|
| LCP: biggest element loads | ≤ 2.5 s | 2.5–4.0 s | > 4.0 s |
| INP: response to a tap | ≤ 200 ms | 200–500 ms | > 500 ms |
| CLS: layout shift score | ≤ 0.1 | 0.1–0.25 | > 0.25 |
The 75th percentile means you can't grade yourself on your best load. If one in four visitors has a slow experience, you fail, and those visitors on older phones are the likeliest to bounce.
Lab data vs. field data: why your Lighthouse score lies to you
A green "95" in PageSpeed Insights and red Core Web Vitals in Search Console can both be true, because they measure different things.
- Repeatable
- Great for debugging
- One simulated device and network
- Not your customers
- Real Chrome users on your site
- Old Androids, spotty LTE, the messy reality
- What Google uses for ranking
- Rolling 28-day window
- so it moves slowly
The honest truth: rankings vs. revenue
The SEO impact is real but modest. Core Web Vitals are a tiebreaker. Relevance and authority still do the heavy lifting, and anyone promising you'll "rocket to page one" by fixing LCP is selling a story.
The conversion impact is where the money is. Speed affects every visit, whether it came from Google, an Instagram ad or an email. The commonly cited numbers are directional, not gospel, but consistent:
Take the pattern to the bank: faster converts better, especially on mobile, on paid traffic and at checkout. Sending ad clicks to a slow page means paying full price and throwing some away at the door.
Where the time goes, and what to fix first
The usual culprits are the same short list: oversized images, render-blocking scripts and CSS, heavy client-side JavaScript, a pile of third-party scripts (chat, pixels, tag managers), fonts that block text, and no CDN. Third-party scripts are the most underestimated; ten "harmless" ones together own your main thread.
- 1Fix the LCP imageCompress, convert to AVIF or WebP, size it for mobile, mark it high-priority.
- 2Set dimensions on everythingImages, video, ad slots and embeds. Can turn CLS green in an afternoon.
- 3Tame your JavaScriptCode-split, defer non-critical code, break up long tasks.
- 4Cut third-party scriptsDelete what doesn't earn its cost; load the rest late or on click.
- 5Self-host and preload fontsUse font-display so text shows immediately.
- 6CDN and faster serverCache at the edge; static generation fixes most slow responses.
Re-measure field data after each change, and stop when you're comfortably green at the 75th percentile.
How a modern stack makes this easier
You can hit good scores on almost any platform, but Next.js makes the fast path the default: image optimization, per-route code-splitting, self-hosted fonts and static or streamed rendering. React Server Components ship less JavaScript in the first place, and the fastest script is the one you never send.
It's not a magic wand. I've audited plenty of slow Next.js sites. The stack removes excuses, not responsibility. The longer argument is in why we build business sites on Next.js.
Frequently asked questions
How much will fixing Core Web Vitals actually increase my sales?
It depends on how slow you are now. Moving from a five-second LCP into the "good" range shows up in conversion rate; going from good to slightly better won't move much.
Is a high PageSpeed Insights score enough?
No. That's a lab test. Track the Core Web Vitals report in Search Console, which uses real-visitor field data.
What's the single most common problem you find?
Images. Oversized files are the top cause of bad LCP, and images without set dimensions are a top cause of bad CLS.
How often should I check these metrics?
Monthly in Search Console, plus any time you add a third-party script, redesign a template or change hosting.
Not sure where you stand? Get a free site speed audit, no pitch, just findings. Deeper fixes are everyday work for our web development team; see typical scope on our pricing page, or reach out for a straight answer.
