Lighthouse Audits for Marketing Sites
What Lighthouse Actually Measures
Lighthouse is Google's open-source auditing tool, built into Chrome DevTools and available via CLI, CI integrations, and PageSpeed Insights (which runs Lighthouse against a lab environment plus real-user CrUX data). It scores five categories — Performance, Accessibility, Best Practices, SEO, and optionally PWA — each from 0-100. For marketing engineers, the Performance and SEO categories matter most because they're the ones directly tied to acquisition and conversion, but Accessibility failures increasingly carry legal and brand risk too (ADA/WCAG lawsuits against marketing sites are common).
Lab Data vs Field Data — A Critical Distinction
Lighthouse in its default mode runs lab data: a single synthetic page load under simulated network and CPU throttling, on a single device profile. This is reproducible and great for debugging, but it is not what your real users experienced. PageSpeed Insights also surfaces field data from CrUX — real Chrome user metrics aggregated over 28 days. A page can score 95 in the Lighthouse lab run and still fail Core Web Vitals in the field if your real traffic mix skews toward low-end Android devices or poor network conditions the lab simulation didn't replicate.
Always check both. Lab data tells you what to fix; field data tells you whether it mattered.
Lighthouse Score Is Not a KPI, It's a Diagnostic
The Audits That Matter Most for Marketing Pages
Within the Performance report, prioritize:
- Render-blocking resources: usually third-party marketing tags loaded synchronously.
- Unused JavaScript/CSS: often an entire component library or analytics SDK loaded for one widget.
- Properly sized images: hero images served at 4x the rendered dimension is extremely common on marketing sites built by non-engineers in a CMS.
- Avoid enormous network payloads: video backgrounds and carousel images are frequent culprits.
- Preconnect to required origins: for GTM, GA4, Meta Pixel, and font CDNs, a <link rel="preconnect"> shaves meaningful time off third-party script negotiation.
# Run Lighthouse headlessly in CI against a staging URL,
# fail the build if performance drops below a budget.
npx lighthouse https://staging.example.com/landing/spring-promo \
--output=json \
--output-path=./lighthouse-report.json \
--chrome-flags="--headless" \
--preset=desktop \
--only-categories=performance,seo,accessibility
# Parse the score and gate the deploy
node -e "
const report = require('./lighthouse-report.json');
const perf = report.categories.performance.score * 100;
console.log('Performance score:', perf);
if (perf < 85) {
console.error('Performance budget breached, failing build.');
process.exit(1);
}
"SEO and Accessibility Audits Marketers Often Ignore
The SEO category in Lighthouse checks surface-level technical hygiene: meta descriptions, valid hreflang, crawlable links, and a robots.txt that isn't blocking the page. These are low-effort fixes with outsized impact since they gate whether the page is indexable at all. Accessibility audits flag missing alt text, insufficient color contrast, and unlabeled form inputs — all of which, beyond compliance risk, also hurt conversion for the meaningful share of users relying on assistive tech or simply viewing your site in bright sunlight on mobile.
Operationalizing Audits Instead of Running Them Ad Hoc
The failure mode most teams hit: someone runs a Lighthouse audit once during a launch review, fixes the obvious issues, and then performance silently regresses over the next six months as new tags, widgets, and content get added. Fix this by wiring Lighthouse CI into your deployment pipeline with a performance budget (see the script above), and separately running scheduled PageSpeed Insights API checks against production URLs weekly to track field data drift over time, not just lab snapshots at deploy time.
Throttling Profiles Can Mislead You
What's Next?
Once you can diagnose speed and SEO issues systematically, the next layer is fixing the single most common root cause on marketing sites: unoptimized images.
I build these systems professionally.
Whether it's a RAG pipeline, analytics migration, or AI workflow — let's talk.