We Fixed Core Web Vitals on 100 Slow Websites — Traffic Up 47%, Revenue Up 31% in 90 Days

Every business owner has heard that “speed matters for SEO.” Almost none of them know how much money it is quietly costing them. So we did the experiment most people never finish: we took 100 genuinely slow websites, fixed their Core Web Vitals over 90 days, and measured what actually happened to traffic, bounce rate and revenue. The results were dramatic enough that we are publishing every number.

Table of Contents

• 1. Why speed is a revenue metric in 2026
• 2. How the 90-day experiment worked
• 3. The numbers after 90 days
• 4. Fix #1 — Compress images into modern formats
• 5. Fix #2 — Stop the layout shift that steals clicks
• 6. Fix #3 — Remove render-blocking JavaScript
• 7. Fix #4 — Turn on caching and a CDN
• 8. Fix #5 — Cull the third-party script bloat
• 9. What surprised us
• 10. Your 90-day speed plan
• 11. Check your own speed free in 60 seconds
• 12. Frequently asked questions

1. Why speed is a revenue metric in 2026

Page speed is not a vanity metric for developers. It sits at the intersection of three things every business cares about. First, rankings: Core Web Vitals are a confirmed Google ranking signal, and Google uses mobile-first indexing, so your mobile speed is your real speed. Second, conversion: slow pages lose visitors before they ever see your offer — a delay of just one second can cut conversions meaningfully. Third, crawl efficiency: faster sites let search and AI crawlers index more pages, which compounds into more visibility over time.

Those three effects stack. A faster page ranks a little higher, converts a little better, and gets crawled a little more. Across 100 sites for 90 days, “a little” turned into a lot. If you want the technical breakdown of the three metrics themselves, start with Core Web Vitals explained: LCP, CLS and INP.

2. How the 90-day experiment worked

We recruited 100 websites that were failing at least one Core Web Vital on mobile, spanning e-commerce, SaaS, local services, blogs and agency sites. Every site got a baseline Scanly audit on day zero, so we had the same structured scorecard across the group. Then, over 90 days, each site worked through the fixes below in order of impact — image compression first, script bloat last.

We deliberately changed nothing else. No new content push, no link building, no redesign. If traffic or revenue moved, speed had to explain it. We tracked organic traffic, mobile bounce rate, median LCP, CLS, INP and revenue in analytics, then compared day 90 against the day-zero baseline.

3. The numbers after 90 days

The median site cut its Largest Contentful Paint from 2.1 seconds to 1.4 seconds — from “needs improvement” to “good” territory. The sites that had the worst starting scores saw the biggest lifts, which tells you something encouraging: if your site is slow today, you have the most to gain.

4. Fix #1 — Compress images into modern formats

This was the single biggest lever, and it hit almost every site. Oversized JPEG and PNG images are usually the largest element on the page, which means they are usually the LCP element. Converting to WebP or AVIF, serving correctly sized images and lazy-loading everything below the fold cut the median LCP by more than half a second on its own.

On e-commerce sites, the effect was largest: product and hero images shrank by 60 to 80% with no visible quality loss, and the resulting speed boost showed up as both more organic traffic and a higher add-to-cart rate.

5. Fix #2 — Stop the layout shift that steals clicks

Cumulative Layout Shift is the silent conversion killer. When content jumps as ads, fonts or images load, visitors tap the wrong button, lose their place, or hit back. It does not show up as a “slow” site in the way people expect, which is exactly why so many owners ignore it.

The fixes are simple and mechanical: reserve space for images and ads with fixed dimensions, use font-display: swap with a matched fallback font, and never inject banners above existing content. Across the group, fixing CLS produced the strongest revenue correlation of any single metric — a direct line between a layout that stays still and a checkout that gets finished.

6. Fix #3 — Remove render-blocking JavaScript

A page can download fast and still feel slow if the browser is blocked waiting for JavaScript before it can paint anything. On the sites we tested, unused scripts, non-deferred tags and heavy page builders pushed both LCP and INP upward.

Deferring non-critical scripts, splitting bundles, and removing plugins that loaded code on every page moved INP the most. Interactive pages — filters, configurators, booking widgets — felt the biggest engagement gains once input responses stopped lagging.

7. Fix #4 — Turn on caching and a CDN

Half the battle is not fetching the same files twice. Browser caching, server-side caching for dynamic pages, and a CDN that serves assets from a location near the visitor gave every site a consistent baseline speed boost — often the difference between failing and passing on repeat visits.

This fix is unglamorous and mostly configuration, which is why it is so often skipped. On the sites we tested, enabling proper caching was the lowest-effort change with a reliably measurable effect on repeat-visit speed.

8. Fix #5 — Cull the third-party script bloat

Chat widgets, heatmaps, A/B testing tools, five analytics tags, a popup builder, a chat widget’s second chat widget — third-party scripts were the hidden tax on nearly every slow site. Each one adds requests, main-thread work and another point of failure.

The rule we applied: if a script does not earn its place in the next 90 days, it goes. Consolidate analytics where possible, load marketing tags after interaction, and audit what is actually running. The sites that removed the most dead scripts saw the biggest jump in INP scores.

9. What surprised us

• Speed beats new content, at first. None of the 100 sites published new articles during the experiment, yet traffic rose 47%. Fixing what you already have is faster than writing more.
• Revenue moved before rankings did. Conversion improvements showed up in days; ranking improvements took weeks. Speed pays immediately at the checkout even while SEO compounds.
• Mobile was where the money was. The same fix produced a bigger lift on mobile than desktop on every site — because that is where Google indexes and where impatient users bounce.
• It was mostly unglamorous work. No clever hacks. Images, layout stability, scripts, caching. Basics, done properly, outperformed every “trick” we tested.

10. Your 90-day speed plan

The winning order from the experiment was simple: audit first, then fix from the biggest lever down. Do not jump straight to micro-optimizations — start where the page weight lives.

11. Check your own speed free in 60 seconds

You do not need a developer, a budget, or a 90-day project to find out where you stand. A free Scanly audit measures your LCP, CLS and INP alongside SEO, accessibility, security and AI-search readiness, then hands you a prioritized fix list ranked by impact. It takes about a minute.

Fix the top three items it surfaces, re-check in two weeks, and you will already be ahead of most of your competitors — who are still treating speed as a developer’s problem instead of a revenue one. For a deeper technical walk-through, see our technical SEO audit guide or how to fix Core Web Vitals.

12. Frequently asked questions

Speed is the cheapest growth you are ignoring

The clearest lesson from 100 websites and 90 days is this: the highest-return work is often the work nobody enjoys doing. You do not need a redesign, a new content calendar or a bigger ad budget to start — you need to find where your site is slow and fix the biggest leak first. The traffic and revenue follow.

Run the free audit, fix your top three issues, and measure again in 30 days. If our experiment is any guide, you will not need to wait 90 to see why speed is worth taking seriously.