Website speed: what to fix before chasing a perfect score.

Understand Core Web Vitals, separate real visitors from lab tests and prioritise practical WordPress and Elementor improvements.

// Good businesses deserve better online.

Useful thinking.
A clear next step.

// THE SHORT ANSWER

Improve website speed by finding what delays the page your customers actually use. Check real-user Core Web Vitals where available, use lab tests to investigate and prioritise images, unnecessary scripts, fonts and delivery. The goal is a responsive, stable customer journey, with evidence that the changes helped.

Three measures. Three different problems.

Core Web Vitals describe loading, responsiveness and visual stability. Their current “good” thresholds are:

  • Largest Contentful Paint (LCP): 2.5 seconds or less. How quickly the largest visible content element appears.
  • Interaction to Next Paint (INP): 200 milliseconds or less. How promptly the page responds visually to user interactions.
  • Cumulative Layout Shift (CLS): 0.1 or less. How much visible content unexpectedly moves.

These thresholds apply at the 75th percentile, assessed separately for mobile and desktop. They describe the experience across visits, not just your fastest test. See Google’s explanation of the thresholds.

Use the distinction when choosing work: a late hero image, a sluggish menu and a button that jumps down the screen are different problems. One optimisation setting is unlikely to explain all three.

Read the two parts of a speed report.

PageSpeed Insights combines field data from real Chrome users with a lab test run through Lighthouse. Its field view covers the previous 28 days; the lab view tests a page under controlled conditions. The results can differ because they describe different experiences.

A new or lightly visited page may have insufficient field data. PageSpeed may show data for the wider website instead, or none at all. “No data” is not a failed assessment. Check whether the report represents the page or the whole origin before interpreting it. Google’s PageSpeed documentation explains these distinctions.

Use the lab findings to locate work worth investigating. Record the URL, device view and test date so comparisons remain useful. Also open the page on a real phone and complete its important actions: a report cannot tell you whether the enquiry questions make sense.

Start with the content people see first.

Inspect the main image before compressing every file in the media library. Is the page delivering a large desktop photograph to a small screen? Is a decorative video doing a job a still image could do? Does the image remain sharp enough after resizing and compression?

Give images an appropriate display size and reserve their space in the layout. Check what loads immediately and what can wait until the reader scrolls. Elementor’s performance settings include image-loading optimisation that prioritises the LCP image and lazy-loads images below the fold.

Check the result on the actual page. A blanket rule that delays every image can work against the first screen you need visitors to see.

Give every plugin and embed a reason to stay.

List the tools that affect a priority page: forms, chat, maps, tracking, sliders, animation add-ons and booking widgets. Identify which serve the customer’s task and which were left behind by an earlier design.

Investigate their measured cost before deciding what to remove. A useful booking tool may justify its weight; an unused animation library may not. Elementor’s speed guidance identifies media, external scripts, plugins, hosting and fonts as areas to examine.

Test changes on a staging copy with a recoverable backup. Disable suspected extras individually, compare the affected page and check the feature that depends on them. Recheck forms, navigation and bookings before deploying. Removing a tool that receives enquiries would defeat the purpose of the exercise.

Simplify fonts and check how the page is delivered.

Audit the typefaces and weights actually used. A site can inherit duplicate font requests from its theme, builder and add-ons. Keep the typography intentional, then check how text appears while fonts load. The fallback should remain readable and should not produce distracting movement when the final font arrives.

Review caching and server response with the hosting provider. Confirm what the host already supplies before adding another optimisation plugin. For a New Zealand audience, test the experience on relevant devices and connections as well as looking at the server configuration.

These are useful questions for an existing WordPress website. Changing platforms is not automatically the first answer to a slow page.

Verify the journey, not just the score.

Prioritise the homepage, important service pages and enquiry path. Fix the largest evidenced issue, then repeat a comparable test and use the page. Check both the initial load and interactions such as opening the mobile menu or submitting a form.

Fresh lab checks can reveal the immediate effect of a change. Field reporting takes time to reflect a new experience because it includes earlier visits. Keep those timelines in mind when evaluating progress.

A perfect Lighthouse score is not a sensible blanket promise. Define the pages, devices, functionality and measures that matter, then report what improved and what remains. Performance belongs within the wider web design brief, alongside useful content and a working enquiry journey. Discuss the pages that need attention.

A few useful answers.

Does “no data” mean my site fails Core Web Vitals?

No. There may not be enough eligible real-user samples to assess the page or website. Use diagnostic tests and direct journey checks while recognising that limitation.

Can an Elementor website be improved without rebuilding it?

Often there is useful work to investigate first: image delivery, unnecessary widgets or scripts, font requests and hosting. The measurements should determine whether the structure itself needs changing.

Should I expect a guaranteed 100 PageSpeed score?

No. Agree practical performance targets and preserve required functionality. A score alone does not demonstrate that a website answers customer questions or produces suitable enquiries.

Put it to work.

Planning a new website or working out what needs to change? Explore web design with Daniel James, or start a conversation about your business.

// The person behind the thinking

DAN SMITH.

Daniel James is Dan Smith: web design and digital growth, based in Christchurch and working with businesses across New Zealand. Strategy, structure and design—with a direct line to the person doing the work.

// Enough reading / A useful next step

GOOD THINKING.
BETTER DOING.

Let’s turn the right ideas into a website that does a better job for your business.