In this guide: The complete website speed checklist: hosting and server, caching and CDN, images and media, themes, plugins and code, plus testing and monitoring.
Use this speed checklist to identify important user journeys, collect a baseline and prioritize verified bottlenecks. It is an implementation workflow, not a report of tests performed on your website or a guarantee of a perfect score.
Bookmark it and run through it every few months.
fetchpriority="high" or preload.font-display: swap.Guides: How to Optimize Images for the Web and Best Image Optimization Plugins.
| Metric | Good target |
|---|---|
| LCP | ≤ 2.5 seconds |
| INP | ≤ 200 ms |
| CLS | ≤ 0.1 |
| TTFB (cached) | As low as possible — ideally well under a second |
| Page weight | As small as practical; watch images and JS |
Don’t chase a perfect score. Chase a site that loads fast for real users on real phones.
Further reading: WordPress Advanced Administration Handbook
Maintain the checklist after meaningful changes and keep security, accessibility and recovery requirements alongside performance goals.
Choose the pages and actions customers actually need: reading an article, opening navigation, submitting a form or completing a checkout. A fast homepage is not enough if the next important action stalls. Record representative routes and the devices or connection conditions relevant to the audience. Keep private data and real payment activity out of routine testing.
The speed checklist should identify the owner of each task and the evidence required to assess it. Mark a check not applicable or unverified when appropriate, rather than assuming every site needs every technology. A brochure website, a store and a membership service have different dynamic behavior. The purpose is a useful plan, not a collection of green boxes unrelated to the business.
Record a baseline with the URL, date, tool, software state, cache condition and device assumptions. Test comparable conditions after changes. A warm cache, different page or different network can explain a score difference. Do not attribute an improvement to one edit when several unrelated components changed together, and do not invent missing measurements to complete a report.
A speed checklist should distinguish laboratory observations from available real-user data. PageSpeed Insights documentation explains its assessment. Field data can reflect a broader period and may be unavailable for some pages. A fresh lab run can help diagnosis without proving that the entire audience immediately experienced the same result.
LCP, INP and CLS describe different aspects of experience. Their good thresholds are not interchangeable with a single overall tool score. Evaluate the relevant field aggregation and device grouping when that data exists. Google’s threshold explanation describes the use of the seventy-fifth percentile. A single favorable visit is not equivalent to passing that population-level assessment.
Use the Core Web Vitals guide with this speed checklist to keep diagnostic labels clear. A low blocking-time observation in a laboratory test does not by itself establish good real-user INP. Record which metric and measurement source support each proposed action rather than describing every delay as the same performance problem.
Review whether a slow initial response comes from the origin, a database operation, an external request or another dependency. Compare cached and relevant uncached behavior. A server location can matter, but a different data center is not a universal cure for expensive application work. Gather evidence before treating a more expensive plan as the first required optimization.
The speed checklist should preserve the workload context: concurrent activity, dynamic routes and actual resource constraints. A page-view count alone does not describe server demand. Ask the host to interpret limits and logs within its support scope. Avoid disabling protections or removing business functions merely to make an uncached demonstration look faster.
Map which layer caches pages, objects and static assets. Confirm who owns each rule and how updates invalidate the relevant output. Avoid overlapping settings whose combined behavior nobody understands. Public content and personalized account information need different treatment. A cache hit is not successful if it serves one user’s private state to another visitor.
Use the cache guide for the implementation review. A speed checklist should validate logged-out and relevant logged-in journeys, cart behavior and content updates where applicable. Cache preloading can also consume resources, so use the supported configuration and monitor the actual effect instead of enabling every option by default.
Confirm which assets and pages pass through the delivery service and which requests still reach the origin directly. A CDN can improve delivery for eligible content without fixing every application delay. Review cache status, relevant geography and the freshness of content after updates. Keep the origin and edge responsibilities documented.
For the speed checklist, test representative locations relevant to the real audience without claiming that one remote probe reproduces every user’s device and network. Do not generate artificial sessions or assume that the observed IP location proves customer demand. Report the actual test conditions and keep delivery performance separate from business traffic-growth claims.
Identify the actual important visible content rather than assuming the same element is always responsible for LCP. Check image dimensions, compression, format support and discovery timing. Lazy loading can help off-screen media but may delay important initial content when applied indiscriminately. Reserve dimensions to reduce unexpected layout movement and keep meaningful alternative text.
The image optimization guide supports this part of the speed checklist. Avoid preloading every image or using a high-priority hint for many competing resources without a diagnosis. Keep useful visual information readable on mobile and desktop. Smaller files are a means to improve the experience, not a reason to remove necessary explanation.
Long work on the main thread can delay a user’s action and its visual response. Inspect the relevant interaction rather than focusing only on initial loading. Google’s INP optimization guidance is a useful reference. A script can be small to download yet expensive to execute, and a delayed script can still interrupt the first important click.
A speed checklist should list third-party widgets with their business owner and purpose. Review whether each is needed on every page and whether a supported lighter approach fits. Do not disable analytics, consent or security functions blindly as a shortcut. Test what changes in the actual user journey and preserve the requirements the site must satisfy.
Inspect the work performed by the complete implementation. A theme name or plugin count does not identify the bottleneck by itself. Use representative diagnostics and check dependencies before removing software. Database cleanup should distinguish unnecessary information from orders, account data and configuration still required by the business.
Connect this speed checklist with the theme evaluation and plugin audit. A custom snippet remains maintained software. Do not move code merely to make the plugin list shorter, and do not approve a destructive cleanup without a suitable recovery plan and an understanding of its actual scope.
Use a test environment where practical and prevent customer notifications, live payments and other production side effects. Change one understood area at a time when that helps diagnosis. Record expected benefits and important risks. An optimization that makes a score higher while breaking a form or keyboard interaction is not a complete improvement.
The staging guide and backup recovery process support this part of the speed checklist. Keep a narrow rollback route and verify the actual saved configuration. A full database restore may discard recent business activity, so it is not an automatic response to every performance change.
Check navigation, text, forms, focus visibility and layout on relevant devices. A desktop tool score does not establish usability on a phone or with a keyboard. Avoid hiding essential content or reducing legibility to make a page appear lighter. Performance and accessibility need to support the same useful journey.
Use the mobile testing workflow alongside the speed checklist. Record observations and the conditions tested rather than claiming universal compatibility from one screenshot. Keep unresolved issues visible to the responsible developer and recheck after the layout or interaction code changes.
Revisit representative pages after releases, new integrations and meaningful content changes. Keep the baseline and change dates so future observations have context. Assign ownership for monitoring and define how a regression is investigated. A perfect score on launch day does not establish that the website will remain fast under every workload.
Use the Search Console guide for correctly scoped reporting and uptime monitoring for availability observations. Keep clicks, sessions, response measurements and business outcomes separate. Do not report an optimization as increased revenue or traffic unless the relevant data actually supports that claim.
The owner selects an article, a category page and a contact journey. The developer records comparable measurements and identifies a large media resource on the article. The team prepares an appropriately sized version and checks visual quality, layout and the relevant timing afterward. This is an example of a workflow, not a measured Hostreta case study.
If the result improves under the documented conditions, retain the evidence and check the other representative pages for unintended effects. If it does not, revisit the diagnosis rather than claiming success because the file is smaller. Keep the next action tied to observations. A speed checklist is most valuable when it helps the team learn which changes matter for the actual site.
Promotional recommendation: consider KAWEDMA’s website development and hosting services for a scoped review of verified bottlenecks. Ask for conditions, evidence, proposed changes and business-journey validation. This recommendation does not guarantee a score, a loading-time reduction or search rankings.
Affiliate disclosure: Hostreta may earn a commission if you purchase through this link. Where a separate workload review supports a hosting change, compare hosting.com’s managed WordPress service. Verify current scope, limits and responsibilities. Infrastructure should follow the evidence, rather than be purchased merely to tick a speed checklist box.
For each proposed action, record the affected journey, supporting observation, expected benefit, implementation effort and validation route. Distinguish a clearly identified oversized image from an uncertain claim that every plugin is slow. Start with changes that solve an observed problem and can be checked safely. Do not rank tasks solely by how dramatic their names sound.
Keep uncertainty visible. A request may be slow because an external service is delayed, while a database operation may be expensive only for a particular administration screen. The appropriate fix depends on that context. Ask the responsible owner to investigate the supported configuration before replacing a whole system. A small, verified adjustment can be more maintainable than a broad rewrite based on incomplete evidence.
A team can define practical limits for representative page types and review additions against them. The budget might cover media, scripts or specific user-experience measurements, but it should be grounded in the site’s needs and actual observations. It is not a universal number copied from another website. Record who approves an exception and how its effect will be assessed.
When a new chat widget, advertisement or video is proposed, identify the business value and measure the complete page with the addition. Consider a supported lighter presentation where it meets the same need. Do not remove essential consent, security or accessibility behavior just to meet a byte target. The speed checklist should preserve the useful service while controlling unnecessary work.
Compare the relevant measurements under comparable conditions and record both improvements and regressions. Verify the business journeys again. If the expected benefit is not observed, keep that result instead of rewriting the record as a success. A failed hypothesis can still improve the team’s understanding of the bottleneck.
At handover, provide the baseline, change record, current owner and remaining checks. Keep sensitive logs or private test details restricted. A future maintainer should be able to explain why a setting exists and how it was validated. That makes the speed checklist a practical operating tool rather than a long list of instructions whose original context has been lost.
We may earn a commission on purchases made through these links, at no extra cost to you.
Don’t miss the chance to participate in the biggest giveaway of this year.