A website that responds promptly and stays stable is easier to use. The effect on enquiries or sales depends on the audience, offer and wider journey. There is no universal conversion percentage that can be attached to every second saved.
Three Core Web Vitals
| Metric | What it measures | Good threshold in Google’s guidance |
|---|---|---|
| LCP — Largest Contentful Paint | When the largest visible content element is rendered | 2.5 seconds or less |
| INP — Interaction to Next Paint | Responsiveness to interactions across the visit | 200 milliseconds or less |
| CLS — Cumulative Layout Shift | Unexpected layout movement | 0.1 or less |
The assessment uses the 75th percentile of visits, with the relevant device grouping. See the Web Vitals guidance for definitions and measurement details. Passing these thresholds is not a guarantee of ranking, conversion or overall accessibility.
Field Data and Lab Tests Answer Different Questions
Field data describes experience from real visits, where enough data is available. A new or low-traffic site may not have sufficient field data. Lab tests run under controlled conditions and help diagnose a particular page or build.
Lighthouse and PageSpeed Insights are useful tools, but scores vary with conditions and are not contracts for every visitor’s experience. Record the device, network, page state and test method. Check important journeys and templates, not only the homepage.
Find the Actual Bottleneck
Inspect the critical content, network requests, image sizes, font loading, script execution and server response. A WordPress site can benefit from appropriate caching and careful plugin selection; an Astro or Next.js site can still be slowed by heavy media, hydration or third-party services.
Compare the implemented system rather than quoting a framework’s supposed universal pass rate. Rendering strategy and dependencies matter more than a technology label alone.
Ten Practical Improvements to Assess
- Serve images at suitable sizes and quality.
- Use responsive image sources and appropriate formats.
- Prioritise the main visible content rather than lazy-loading it indiscriminately.
- Defer below-the-fold media when appropriate.
- Reserve space for images, embeds and changing content.
- Reduce unnecessary JavaScript and long main-thread tasks.
- Review third-party analytics, chat and marketing scripts.
- Load only required font files and plan fallback behaviour.
- Configure caching and server delivery for the actual content.
- Retest the user task, including interaction and error states, after each material change.
An optimisation should preserve required functionality and access. Hiding important content or removing keyboard behaviour to improve a score is not a successful result.
Set Acceptance and Monitoring
Agree budgets and targets with their measurement conditions. Record what was tested locally and what will need real-user monitoring after release. Performance can change when editors upload new media or integrations add scripts, so assign ongoing responsibility.
The work required depends on the existing implementation and causes. A quote should follow diagnosis; there is no universal price for reaching a particular score.
Working with Q-Dev
Q-Dev is QA10’s software house offering. Scope, price and schedule are agreed for each project. Four gates cover design, MVP, deployment, and training with the start of maintenance. Each gate has a review meeting and an agreed corrections document; after approval, two further rounds of changes to completed work are available at that gate. Changes may affect timing; price changes only outside the contract scope. The first six months of maintenance are included. Code rights transfer after full project payment.
Describe your website project.
Frequently Asked Questions
Does a Lighthouse score prove real-user performance?
It is a controlled diagnostic result, not a measurement of every visit. Combine lab tests with field data where available and record the test conditions.
Which Core Web Vitals thresholds are considered good?
Google’s guidance uses LCP at or below 2.5 seconds, INP at or below 200 milliseconds and CLS at or below 0.1, assessed at the 75th percentile of visits.
Will a faster site guarantee more sales?
No. It can remove friction, but outcomes also depend on the offer, audience, acquisition and sales process. Measure the relevant business journey.