Website performance: load faster, convert better

Web Development

Website performance: load faster, convert better

Slow pages cost conversions and rankings. This article explains the main performance levers, from images and caching to Core Web Vitals.

6 min readBy Andre Schild and Albert SchaperAuf Deutsch lesen

Why speed is business

Users don't wait. Slow pages mean higher bounce rates, fewer enquiries and weaker SEO signals. Google treats Core Web Vitals as a ranking factor, so performance is not just a developer topic, but marketing and sales.

You don't need to cite Amazon studies: measure yourself, PageSpeed Insights, Search Console (Core Web Vitals report) or Lighthouse in your build pipeline.

Core Web Vitals: the three metrics

MetricWhat it measuresTarget (rough)
LCP (Largest Contentful Paint)When main content becomes visibleunder 2.5 s
INP (Interaction to Next Paint)Response to interactionsunder 200 ms
CLS (Cumulative Layout Shift)Layout shifts while loadingunder 0.1

INP replaced FID; if your tools still show FID, switch to INP.

Field vs. lab: PageSpeed Insights and Lighthouse are lab snapshots. For rankings, Google mainly uses field data (CrUX) over about 28 days. Treat the Search Console Core Web Vitals report as truth, lab tools as diagnosis. After a fix, field data often needs weeks before “Good” appears.

Impact matrix: what pays off first

LeverEffortTypical impactWhen first
Hero/list images (WebP, width)lowhigh (LCP)always on marketing sites
Defer/remove third-party scriptslow-mediumhigh (INP, often LCP)chat, tag manager, embeds
width/height or aspect-ratiolowmedium (CLS)banners, galleries, embeds
Fonts (font-display, subsetting)mediummedium (LCP/CLS)custom fonts without swap
CDN / hosting TTFBmedium-highmediumslow TTFB, global audience
Large refactor (CSR → SSR)highhighwhen lab and field stay red

Rule: one red metric first, not everything in parallel. With several reds, usually LCP (images) before CLS before INP.

Images: usually the biggest lever

Images are often the heaviest part of the page. Priority:

  1. Modern formats: WebP or AVIF, JPEG/PNG only as fallback
  2. Right size: srcset/sizes, mobile devices don't need 4K hero images
  3. Lazy loading: below-the-fold with loading="lazy", hero image eager
  4. Compression: 80-85% quality is usually enough; remove EXIF metadata
  5. SVG for icons and simple graphics

A relaunch with optimised images alone can noticeably improve LCP, often more than CSS tricks.

Code and delivery

Minification: minify CSS and JavaScript in production; remove unused CSS (purging) where your stack allows.

Compression: Brotli or GZIP on the server for HTML, CSS, JS.

Code splitting: with React/Next.js, load only what the page needs (see Next.js vs. React).

Reduce render-blocking: inline or load critical CSS early; defer/async non-critical JS.

Third-party scripts: analytics, chat widgets, tag managers, every script costs. Only load what you actually use; don't let consent banners and tracking block rendering excessively.

External scripts affect both load time and attack surface. The web security guide adds checks for CSP, updates and safe dependencies.

Caching and CDN

Browser cache: static assets (CSS, JS, fonts, images) with long cache headers; cache HTML for shorter periods. Versioned filenames for cache busting.

Server cache: for dynamic sites, Redis/Memcached for expensive queries.

CDN: for international visitors or static assets, Cloudflare, CloudFront, etc. German edge locations also pay off with a broad audience.

Returning visitors benefit most, first visit often remains the bottleneck.

Checklist for the next sprint

  1. PageSpeed Insights for homepage + one subpage
  2. Identify the largest image and re-export (WebP, correct width)
  3. Audit third-party scripts, what can go or be deferred?
  4. Search Console: prioritise pages rated "poor" on Core Web Vitals
  5. Measure again after deploy, not just locally

Performance and SEO go together; for local businesses also check responsive design and technical foundations.

Measurement workflow for SMBs and service businesses

Many teams in Hannover and the region optimise “by feel” without before/after. Better:

  1. Baseline: document PageSpeed Insights for homepage + key service page (LCP, INP, CLS)
  2. Bottleneck: identify largest image, heaviest script or slowest template
  3. One change: only one lever per deploy (e.g. hero image to WebP + correct width)
  4. Re-test: same URL, same tool, note the result
  5. Search Console: prioritise URLs rated "Poor" for Core Web Vitals

That shows whether a relaunch or targeted optimisation is enough before budget goes to the wrong lever.

When to DIY, when to hire a developer

DIY or agency support in a sprint: image exports, lazy loading, redundant plugins/scripts, cache-header checks, Search Console priorities.

Developer needed: INP from heavy JavaScript bundles, layout shifts in templates, hosting/CDN rebuild, framework migration, consent and tag architecture.

For local B2B and service sites in Hannover, a focused sprint on homepage and service page is often enough, not a month-long “performance in general” project.

FAQ: website performance

Do Core Web Vitals really affect rankings?

Yes, as a quality signal. For local and B2B sites they often matter indirectly through better UX and conversion. Field data in Search Console counts more than a single Lighthouse score.

What helps more: CDN or image optimisation?

Usually image optimisation first, especially hero graphics and uncompressed uploads. CDN helps most with distributed visitors and static assets. If TTFB stays above roughly 800 ms, check hosting/CDN in parallel.

How often should I measure performance?

After every major deploy and at least monthly for top 5 pages. Use Search Console Core Web Vitals as an early warning system. Lab tests before and after every lever; judge field data only after 2-4 weeks.

Can I treat performance and SEO separately?

Only partly. Slow pages hurt rankings, ad quality and enquiries at the same time. Performance is therefore a marketing, sales and SEO topic.

Is a one-off PageSpeed score of 90 enough?

No. Lab scores fluctuate and do not cover every user. Aim for “Good” on LCP/INP/CLS in field data for important URLs, plus a conversion path (form, call button) that responds quickly on mobile.

Key takeaways

  • Core Web Vitals (LCP, INP, CLS) are measurable and SEO-relevant
  • Image optimisation often delivers the biggest gain
  • Third-party scripts are an underestimated drag
  • Caching helps most for returning visitors
  • Measure before and after every change, don't guess

Product pages

PerformanceCore Web VitalsSpeedSEO