TL;DR
- Measure representative pages with the real forms, tracking and media installed, not a clean staging homepage.
- Check the first-screen element, images, fonts, scripts and conversion forms separately, and give each finding an owner.
- Retest the changed pages and schedule a new check whenever a marketing tool is added.
Performance work is easier to manage when each check has an owner and an expected result. “Optimise the site” is too broad to work as a launch task.
This checklist is written for a company marketing site. Run it on representative pages with the real forms, tracking and media installed. A clean staging homepage isn’t the finished website.
Establish a baseline
Record the pages tested, the date, the device profile and the conditions. Include the homepage, a CMS detail page, a resource listing and any important conversion page the site has.
Keep repeatable lab measurements alongside whatever field data is available. For a new site without field history, say so. Don’t describe a lab result as the measured experience of every visitor. The reading guide Reading a Lighthouse report without chasing the score explains how to interpret those numbers.
First-screen content
- Identify the likely Largest Contentful Paint element.
- Check that it can be discovered without waiting on an unnecessary script.
- Keep essential content visible even if a decorative animation fails to start.
- Avoid lazy-loading the main image visitors need straight away.
- Check server response and redirects before assuming the image is the delay.
If the hero is text, font loading and reveal effects deserve attention. If it’s an image, inspect the chosen file and its request timing. The image side is covered in Responsive image variants in Webflow, and the sizes value.
Images and video
Check the actual downloads at phone and desktop widths. Verify responsive candidates where they apply, rather than assuming every upload has them.
Reserve space using dimensions or a suitable aspect ratio. Review images in rich text, custom embeds and CSS backgrounds separately. Make sure compression hasn’t made diagrams or screenshots unreadable.
For video, confirm which files load before scrolling. Use deliberate posters, sensible exports and a playback policy that handles reduced motion and failure. Pausing an offscreen video doesn’t necessarily stop the download it already started. The video-specific checks are in Background videos without the bandwidth bill.
Fonts and styles
List the font files the page actually uses. Remove families and weights nobody needs. Check that preloads match resources the page really uses, and that the fallback layout stays readable. The fonts article covers this in more detail.
Review shared styles before adding another page-specific stylesheet. Test first visits and later navigation, because a change that helps one can cost the other.
Don’t optimise around an empty content block. Longer real headings and translated copy can expose wrapping and layout problems that placeholder text hides.
JavaScript and external tools
Inventory the scripts. Each one should have a purpose, an owner and a loading location. Check for duplicate libraries and integrations left behind after a tool was replaced. The custom code snippets guide explains how to keep that inventory from drifting.
Ask which tools are needed before any interaction and which can wait. Booking widgets and below-the-fold effects may have different needs from consent handling or lead measurement.
Measure responsiveness during real use. Open the menu, filter a list and submit a form. A quick initial load doesn’t guarantee a smooth interaction.
Conversion pages
Test forms from start to finish. Confirm the lead reaches the expected system, important fields are mapped, success and error messages work and keyboard focus is sensible.
Check the same page after accepting and after declining the relevant consent options. Verify the analytics events you expect, without sending repeated fake leads into a live sales process.
If a performance change delays a third-party form, make the waiting state clear and keep an alternative contact route. A lighter page that stops enquiries from arriving hasn’t been optimised successfully.
Turn findings into priorities
| Finding | Useful next action |
|---|---|
| Oversized image | Replace it or correct its responsive delivery; compare transferred bytes |
| Hidden hero waiting for an animation | Make primary content visible independently of the effect |
| Duplicate script | Identify the owner and remove one installation safely |
| Moving form or image | Reserve space and retest the affected layout |
| Slow input response | Profile the interaction and find the blocking work |
Retest the changed pages, keep the evidence and schedule a check whenever new marketing tools are added. Sites change after launch, and the performance baseline should stay useful when they do. If you want that work done for you, the website audit service covers it.