Why a "Good" Lighthouse Score is Lying to You: A 53-Point Wake-Up Call
A few weeks ago, a client came to WL Tech saying their site "felt slow." They had already done everything "by the book": optimized their images, installed a premium caching plugin, and run a standard Lighthouse test. The score came back at 72 as the saying goes, "not great, not terrible."
But when I ran my full audit stack on it, I uncovered 53 separate performance and SEO issues that their previous checks completely missed.
Here is exactly what standard automated tools miss, and how to actually find them.
1. Third-Party Script Bloat (17 scripts, 4 critical)
The client had a massive stack of plugins and third-party tools running in the background: analytics, chat widgets, and tracking pixels. Five of them were loading render-blocking scripts on every single page — including text-only pages like their privacy policy and terms of service.
-
The Fix: I implemented strategic resource hinting (
async,defer,preconnect) and deferred non-critical scripts until well afterDOMContentLoaded. - The Result: Largest Contentful Paint (LCP) dropped by 1.8 seconds.
2. Image srcset Gaps (12 instances)
Their images were technically "optimised," and they even used srcset. The problem? The breakpoints were desktop-first. On mobile devices, the browser was forced to download images four times wider than the actual layout container.
-
The Fix: Rewrote the
sizesattribute to accurately match their mobile CSS layout. - The Result: Total page weight dropped by 40%.
3. Core Web Vitals Gaps That Lighthouse Ignores
Lighthouse flagged the obvious metrics like LCP and CLS, but it completely ignored the silent killers:
- INP (Interaction to Next Paint): Their primary CTA button took a massive 680ms to respond. It felt sluggish to users, but Lighthouse didn't flag it as "poor."
- TTFB (Time to First Byte): It took 1.2 seconds just to get the first byte back, despite being on a "fast" hosting plan. The culprit? A blocking API call firing in their theme before anything could render.
- Unused CSS: A massive 180KB framework was loading on every page. Only 23KB of it was actually being utilised.
4. SEO Issues Masquerading as Performance
Their product and local business schema markup was broken. Google couldn't parse critical data like price, availability, or service areas. While this isn't technically a "performance" metric in a speed test, it directly impacts how fast they acquire customers through organic search.
The Ultimate Fix
Instead of a generic automated printout, I delivered a comprehensive Developer Report—a 12-page breakdown mapping out every single issue, its exact business impact, and the precise fix required.
Their developer implemented the top 5 highest-impact issues over a single weekend.
The ultimate payoff? Their mobile score jumped from 72 → 94.
Building This at Scale
To make this a sustainable business model without spending hours manually auditing every single site, I built a dedicated backend infrastructure:
- An Automated Audit Pipeline: A multi-tool stack combining Lighthouse, WebPageTest, and custom scrapers.
- A Tokenized Report System: Dynamically generates these high-value Developer Reports in about 15 minutes.
- A Flat-Rate Pricing Model: No messy discovery calls or hidden fees. Just $150 for the comprehensive report, or $250 for a Quick Fix on the top 3 critical issues.
No monthly retainers. No contract lock-ins. Just raw data and the exact fixes you need.
Want to see what you're missing? If you're a developer auditing client sites, or a small business owner curious about your actual performance, run your site through the free audit tool at wltech.pro to see what standard tools are hiding from you.
Christopher Welsh runs WL Tech—specialising in website performance and technical SEO for small businesses. Based in the Scottish Borders, serving the UK, EU, US, and international clients.