PageSpeed optimisation service
Chasing a number is not enough — you need changes that improve real load time, SEO, and conversions. We translate Lighthouse audits into work your team can ship, with before and after measurement for every change.
A low PageSpeed score is a warning sign: heavy images, slow servers, or JavaScript blocking the main thread before your content can render.
Aligned with Google PageSpeed Insights and real-user field data where available.
Google PageSpeed Insights scores your site from 0 to 100 based on how quickly it loads and becomes usable on mobile. A score below 50 sits in the red band — a direct signal that real users are waiting too long before they can read, scroll, or interact. Most sites in this range are not failing because of one big problem; they are failing because of five or six overlapping issues that each add a few hundred milliseconds.
The score is calculated from lab metrics — Largest Contentful Paint, Interaction to Next Paint, Cumulative Layout Shift, and others — that together reflect how a page feels to a visitor on a mid-range mobile device on a standard mobile connection. A score of 88 on desktop but 34 on mobile is common, because desktop hides the cost of heavy JavaScript and large images behind faster CPU and network speeds. Google primarily uses mobile performance when ranking your pages.
Search engines and ad platforms reward fast experiences. Pages that load slowly see higher bounce rates before the content appears, lower conversion rates from visitors who stay, and weaker organic rankings over time. A one-second improvement in mobile load time has been measured to lift conversions by 7–10% for ecommerce stores. For lead generation sites, the same improvement reduces form abandonment and cost per lead from paid traffic.
The gap between a score of 35 and 80 is usually closed by the same set of fixes: compressing images and serving them in modern formats, deferring render-blocking JavaScript, configuring caching correctly, and eliminating layout shift. These are well-understood problems with well-understood solutions — the challenge is identifying which issues affect your specific site and implementing fixes that survive future content and plugin changes.
35→80
typical mobile score range
Sites starting between 25 and 45 on mobile PageSpeed commonly reach 70–85 after targeted image, script, and caching improvements — no redesign required.
7–10%
conversion lift per second saved
A one-second improvement in mobile load time produces measurable revenue uplift for ecommerce stores, according to Google's research.
53%
of mobile users leave at 3 seconds
More than half of mobile visitors abandon a page that takes longer than three seconds to load. Most unoptimised sites exceed that threshold on mobile.
Every fix maps to a specific Lighthouse audit so progress is measurable before and after.
Hero and product images are converted to WebP format, resized to match the largest viewport they are served at, and given explicit width and height attributes to prevent layout shift. Preload hints are added for the LCP image so the browser fetches it at the highest priority before discovering it in the HTML. On most sites, image work alone moves LCP from failing to passing.
We configure browser caching headers so static assets are served from local cache on repeat visits rather than re-downloaded. A CDN is configured to serve assets from edge locations close to your visitors, reducing round-trip time for users outside your server's region. Stale-while-revalidate policies allow cached pages to be served instantly while fresh versions are fetched in the background.
We audit every stylesheet and script that loads in the document head and blocks the browser from painting the page. Non-critical CSS is deferred or loaded asynchronously. Above-fold critical styles are inlined directly into the HTML so the first paint does not wait for a full stylesheet download. The result is a measurable reduction in First Contentful Paint and Total Blocking Time.
Third-party scripts — analytics tags, chat widgets, marketing pixels, A/B testing tools — are profiled individually for their network and execution cost. Non-critical scripts are deferred until after the page is interactive. Scripts that are never used are removed. For scripts that cannot be deferred, we look for lighter alternatives or consolidation through a tag manager. This directly reduces Total Blocking Time, the second-highest-weighted PageSpeed metric.
We address each of the three Core Web Vitals directly: LCP by optimising hero images and reducing TTFB; INP by profiling and reducing main-thread JavaScript tasks that delay tap responses; and CLS by adding explicit dimensions to images and embeds, stabilising font loading, and preventing late-injected banners from shifting content. Each metric is measured in both lab and field data.
Images and iframes that appear below the initial viewport are given native lazy loading attributes so the browser defers their download until the visitor scrolls toward them. This reduces the initial page weight, improves Speed Index, and allows the browser to prioritise the LCP element. Third-party embeds — maps, videos, review widgets — are replaced with click-to-load facades that defer the full embed until user interaction.
We audit every font family and weight loaded on each page and reduce them to the minimum required by the design. Google Fonts are either self-hosted for cache efficiency or given preconnect hints to reduce connection time. The most-used font weight is preloaded so above-fold text renders immediately. Font-display swap is applied to prevent invisible text during load, eliminating a common source of CLS.
Time to First Byte is measured across key templates. When TTFB exceeds 600ms, we diagnose the cause: slow PHP execution, missing object caching, insufficient database indexing, or hosting limitations. We implement server-side page caching and object caching where available, and provide a clear recommendation when hosting itself is the ceiling preventing further improvement.
PageSpeed Insights uses Lighthouse to generate a weighted score from six metrics. Largest Contentful Paint and Total Blocking Time carry the most weight — together accounting for roughly 60% of the total score. LCP measures when the largest visible element on screen finishes loading: usually the hero image or a large heading. If that element loads in under 2.5 seconds you pass; at four seconds you are in the red. Total Blocking Time measures how long the main thread is blocked by JavaScript before the page can respond to user input. A single third-party marketing script loading synchronously can add 500ms of blocking time on its own.
The practical implication is that not all optimisations move the score equally. Lazy-loading images below the fold improves Speed Index and reduces page weight, but it does not move LCP because LCP measures above-fold content. Minifying CSS reduces file size but rarely moves Total Blocking Time because CSS is not the primary bottleneck on most sites. The changes that move scores most reliably are: serving the hero image as WebP with correct dimensions and a preload hint, deferring non-critical JavaScript (especially third-party tags), and ensuring the server responds in under 600ms. Those three changes address the two highest-weighted metrics directly.
Field data versus lab data is another important distinction. PageSpeed Insights shows both: lab data from Lighthouse run in a controlled environment, and field data from real users captured via Google's Chrome User Experience Report. The two can diverge significantly — a site that scores 72 in the lab may show Needs Improvement in field data if real users have slower devices or the page behaves differently for logged-in visitors. Lab data is actionable and reproducible; field data is what affects your organic ranking signal in Google Search Console.
Representative outcomes from optimization projects. Your results depend on stack and traffic.
SaaS marketing site
Before
Mobile PSI 29
After
Mobile PSI 84
Hero WebP conversion and critical CSS inlining
Local services brand
Before
Desktop 52
After
Desktop 91
CDN configuration and cache headers
Ecommerce landing
Before
FCP 3.1s
After
FCP 1.2s
JavaScript budget enforced and third-party scripts deferred
We run a deep speed and Core Web Vitals audit on your key templates — home, collection, product, and checkout where relevant. Every issue is ranked by its impact on real user experience and revenue.
We implement prioritised fixes: images, scripts, caching, and theme or app bloat — focused on revenue pages first. Each change is tested in staging and measured against a baseline.
You get ongoing checks so regressions from new apps, themes, or campaigns are caught before they hurt SEO and sales. Alerts fire when scores drop below your target threshold.
Explore our speed optimization services
Straight answers about timelines, risk, and what to expect.
Core Web Vitals fix service to improve LCP, INP, and CLS scores. We diagnose failing metrics and implement changes that help SEO and real-user experience.
Learn more →Professional Shopify speed optimisation to improve Core Web Vitals, reduce load time, and recover lost conversions. Free audit available for all store sizes.
Learn more →Expert WordPress speed optimisation to fix slow loading, improve Core Web Vitals, and boost SEO. Get a clear, prioritised fix plan for your WordPress site.
Learn more →Run a free PageSpeed check with Loadzen. Get issues, suggestions, and platform-specific next steps.