InsightsEngineering & Case Study
How We Achieve 100/100 Lighthouse Scores for Client Websites: A Technical Case Study

How We Achieve 100/100 Lighthouse Scores for Client Websites: A Technical Case Study

Written byCazim
Published2026-04-10

How We Achieve 100/100 Lighthouse Scores for Client Websites: A Technical Case Study

At WebCreates, we have a saying: "If it's not sub-second, it's not finished." In 2026, the digital landscape is unforgiving. A website with a Lighthouse score in the 70s or 80s is no longer just "slightly slow"—it is a liability. It bleeds marketing budget, kills conversion rates, and is systematically penalized by Google’s Search Generative Experience (SGE).

Achieving a perfect 100/100 Lighthouse score across Performance, Accessibility, Best Practices, and SEO is not an accident. It is not the result of a single "magic" plugin. It is the outcome of a rigorous, multi-layered engineering process.

In this 3,000-word case study, we pull back the curtain on our internal workflow. We will analyze a real-world transformation of a legacy enterprise site into a high-performance Next.js 15 powerhouse.


1. The Starting Point: Diagnosing the "Performance Decay"

Our client, a mid-market e-commerce brand in the renewable energy sector, came to us with a typical problem. Their WordPress/Liquid hybrid site was "functional" but failing every modern metric.

The Baseline Metrics:

  • Lighthouse Performance Score: 42/100 (Mobile)
  • Largest Contentful Paint (LCP): 4.8s
  • Interaction to Next Paint (INP): 340ms (Failing)
  • Total Blocking Time (TBT): 1.2s

The site felt heavy. Users on mobile devices experienced "jank" (visual stutters), and the bounce rate on Facebook Ads was hovering at 65 percent. The cost-of-slow-website was costing them an estimated 15,000 EUR in lost revenue every month.


2. Phase 1: Architectural Deconstruction

The first step in our global performance strategy is moving away from the monolith. We replaced the heavy PHP/Liquid backend with a decoupled architecture.

Why Next.js 15?

We chose Next.js 15 for its Server Components (RSC). By moving the majority of the application logic to the server, we drastically reduced the amount of JavaScript the client's browser had to execute.

The Result: Our baseline bundle size dropped from 450kb of "mandatory" JS to less than 60kb. This immediately solved the "Main Thread Blocking" issue that kills most Lighthouse scores.


3. Phase 2: Asset Orchestration (The LCP Killer)

Images and fonts are usually the primary reason for low performance scores. At WebCreates, we treat assets as dynamic data, not static files.

1. Zero-Layout Shift Images

We utilized the Next.js Image component with a twist. By implementing "Blurry Placeholders" and explicit aspect ratios, we ensured a Cumulative Layout Shift (CLS) of exactly 0.000.

2. AVIF at the Edge

Using Vercel’s built-in image optimization, we automatically serve AVIF formats to browsers that support them (like Chrome and Safari in 2026). This reduced image payloads by 60 percent without any perceptible loss in quality.

3. Font Loading Strategy

We moved away from Google Fonts (which require an extra DNS handshake) and shifted to Local Self-Hosted Fonts. Using the next/font API, we ensured that fonts were preloaded and swapped instantly using the size-adjust property to prevent any visual jumps.


4. Phase 3: The Third-Party "Sanitarium"

The #1 reason sites fail the Best Practices and Performance audits is unoptimized third-party scripts (GA4, Meta Pixel, Hotjar).

The Partytown Integration

We don't load tracking scripts on the main thread. Instead, we use Partytown to offload these scripts to a Web Worker.

  • The Logic: Your marketing tools still collect data, but they don't "steal" CPU cycles from the user.
  • The Score Impact: This single move often improves the "Total Blocking Time" (TBT) by 400-600ms.

5. Phase 4: Edge-First Deployment Infrastructure

A site is only as fast as the server that delivers it. We utilized Vercel deployment to place our client’s site as close to their customers as possible.

Time to First Byte (TTFB) Optimization

By using Partial Prerendering (PPR), we serve the static shell of the page from the Edge in under 80ms. The dynamic content (user login, cart state) streams in asynchronously. For a user in Sarajevo or Berlin, the page feels "instant" because the initial HTML is served from a server in their own region.


6. Phase 5: Accessibility and Best Practices (The 100/100 Sweep)

Lighthouse isn't just about speed; it's about quality.

Semantic HTML and ARIA

We audited every component for keyboard navigability and screen-reader compatibility. In 2026, accessibility is a major SEO signal. By using Radix UI primitives, we ensured that complex elements like accordions and modals were fully compliant out of the box.

Secure by Default

We implemented a strict Content Security Policy (CSP) and ensured all headers (HSTS, X-Content-Type-Options) were configured correctly. This pushed the "Best Practices" score to a perfect 100.


7. Measuring the Results: The New Baseline

After 6 weeks of engineering, we re-ran the Lighthouse audit.

The New Metrics:

  • Performance: 100/100
  • Accessibility: 100/100
  • Best Practices: 100/100
  • SEO: 100/100
  • LCP: 0.8s (Down from 4.8s)
  • INP: 38ms (Down from 340ms)

8. Business Impact: Beyond the Score

While 100/100 looks great on a report, the real success was in the client’s analytics.

  1. Mobile Conversion Rate: Increased by 28 percent within the first month.
  2. Organic Search Traffic: Impressions grew by 42 percent as Google rewarded the improved technical health.
  3. Ad Spend Efficiency: Cost Per Acquisition (CPA) dropped by 15 percent due to a better landing page experience score.

9. Continuous Performance: The CI/CD Pipeline

A 100/100 score is a snapshot in time. To prevent "performance rot," we integrated Lighthouse CI into our Vercel workflow. Every time a developer pushes code, a bot runs a performance test. If the score drops below 98, the build is blocked. This ensures that the Next.js speed we worked so hard to achieve remains permanent.


10. Conclusion: Performance is an Engineering Choice

Achieving a 100/100 Lighthouse score is a signal of respect for your users. It says that you value their time and their data.

At WebCreates, we have turned this process into a science. We don't just build websites; we build high-performance assets that dominate their niche. Whether you are a SaaS founder or a global e-commerce brand, the "Sub-Second Standard" is your ultimate competitive advantage.

Ready for Your Transformation?

Your business deserves to be in the 100/100 club. Let’s look at your current site and map out a path to technical excellence.

Request Your Performance Case Study Audit


About the Author

Cazim is the founder of WebCreates. He is a lead web engineer specializing in Next.js 15 and high-performance Edge infrastructure. He has helped dozens of brands transition from legacy "slow" web to the modern "instant" web.

Want a site this fast?

We specialize in building high-performance Next.js sites that rank. Lets discuss your project today.

Book a Free Consultation
Live Chat
Ask anything