Consider a Webflow marketing site that shows Performance 99 in Lighthouse while Search Console lists its URLs as failing Core Web Vitals. That is a hypothetical, but it is a common shape, and both reports can be correct. Your first action is to open the page in Chrome DevTools, record a reload in the Performance panel and find the LCP marker. Everything else in this post depends on knowing which element that is.
The advice for Webflow speed usually stops at "compress your images". The scoring model decides the outcome, so this is the order I would work in and the numbers that count. All figures below were checked against Google and Webflow pages on 30 September 2026.
What Makes a Webflow Site Slow?
A Webflow site is usually slow because of what gets added to it: oversized hero images, a lazy-loaded hero and third-party scripts pasted into custom code, with images and scripts weighted differently in the score.
Lighthouse scores Total Blocking Time at 30% of the performance number, Largest Contentful Paint and Cumulative Layout Shift at 25% each, and First Contentful Paint and Speed Index at 10% each, according to Chrome's Lighthouse scoring documentation. Third-party JavaScript adds main-thread work, which is what blocking time measures, so this suggests an unaudited chat widget can cost more score than an uncompressed image. Treat that as an inference to test on your own site, since the mix differs on every page.
Picture a small-business site where the team has added a chat widget, a heatmap recorder, two ad pixels and a web font from another domain, all in site-wide custom code. That is a labelled hypothetical, but the fix is real: open the Network panel, sort by transfer size and ask who requested each third-party domain. Delete the ones nobody can justify, and defer or load on interaction the rest.
Does a Lighthouse Score of 99 Mean Google Sees a Fast Site?
Not necessarily. Core Web Vitals are judged on field data from real users at the 75th percentile, while Lighthouse is one simulated lab run, so a green ring and a failing report can both be true.
web.dev's Core Web Vitals guide sets the thresholds: LCP within 2.5 seconds, INP at 200 milliseconds or less and CLS at 0.1 or less, measured at the 75th percentile of page loads, segmented across mobile and desktop. The same page says field measurement captures the complete picture that lab tests cannot. The 75th percentile is your slowest quarter of real visits, on real phones and real networks, which is why a lab run on a fast machine flatters you.
The top of the scoring curve is also compressed. Chrome's documentation says taking a score from 99 to 100 needs about the same metric improvement as taking one from 90 to 94, and that an LCP of about 1,220ms already maps to 99. Two sites can therefore both report 99 with different real hero timings. This suggests chasing the last point buys little. Use the score as a diagnostic, read the opportunities list, and judge success on field data.
A one-second load is a goal you can set for yourself, but no threshold Google publishes uses it. The published bar for LCP is 2.5 seconds at the 75th percentile, so that is the number that decides whether you pass.
How Do You Fix a Webflow Hero Image That Hurts LCP?
Set the hero image, which is usually the LCP element, to load eagerly and give it a high fetch priority, because Webflow makes new images lazy-load by default and lazy loading delays the element being timed.
Webflow made all new images lazy-load by default in August 2020. The update does not affect existing images, lets you override the setting per image, and notes that lazy loading covers inline images only, not background images. That is the right default for below-fold content. For the hero, web.dev's LCP optimization guide warns to never lazy-load your LCP image, because the browser waits until layout confirms the image is in the viewport before fetching it.
Fetch priority is the second half. web.dev's Fetch Priority guide reports Google Flights improving LCP from 2.6 seconds to 1.9 seconds in a test after boosting the LCP image with fetchpriority="high", and an Oodle app experiment where lowering the priority of offscreen carousel images cut page load by 2 seconds. Those are results from other sites, so measure your own.
| Image role | Load setting | Extra attribute | Reasoning |
|---|---|---|---|
| Hero / LCP element | Eager | fetchpriority="high" | Starts early instead of waiting for layout |
| Above-fold logo | Eager | none | Small and needed immediately |
| Below-fold section images | Lazy (default) | none | The default is correct here |
| Offscreen carousel frames | Lazy | fetchpriority="low" | Yields bandwidth to the LCP element |
| CSS background hero | No image setting exists | Preload it | Lazy loading does not cover background images |
Webflow's element settings allow custom attributes, but confirm in the published HTML that fetchpriority actually appears on the image.
Where Does Webflow's Image Pipeline Stop Working?
Responsive variants are generated only for images uploaded directly in Webflow, for example in the Assets panel or a CMS image field. Webflow's responsive images help page states that images added by CSV import or the API do not get variants, and that the feature does not apply to background images.
For an original wider than 3200px, the same page says Webflow builds seven variants at 3200, 2600, 2000, 1600, 1080, 800 and 500 pixels, and uploads must stay under 4MB. Consider a blog collection filled by CSV import with image URLs: every image would be served at its original size to every device. The site would look fine in the Designer and struggle on a phone. Check one CMS image on the live page and see whether its URL carries a variant size.
Webflow has supported AVIF uploads and a conversion tool since July 2024, so re-encoding heavy JPEG and PNG assets is a built-in step.
Can You Remove Webflow's jQuery?
Nothing in the sources reviewed for this post shows a dashboard switch to disable it. A Webflow Wishlist request, filed in March 2024 with 15 votes, asks for one and says jQuery loads on every Webflow site. A May 2025 comment there calls it the biggest performance blocker left on that person's site.
Check the Network tab of your published page to confirm what loads. All your controllable savings then come from your own scripts: defer them, move them to footer custom code, load chat widgets on first interaction and delete tags from campaigns that ended long ago.
What Order Should You Fix Things In?
Fix the LCP element first, third-party scripts second and everything else after, because that order follows the metric weights and the field-data thresholds instead of the length of the audit list.
- Identify the actual LCP element in the DevTools Performance panel. It is usually the hero image and sometimes a heading in a web font.
- Set that element to Eager with a high fetch priority. For a background image, preload it in page-level custom code instead.
- Audit third-party scripts. Anything not needed for first paint moves to the footer with
deferor loads on first interaction. - Convert heavy assets to AVIF or WebP with the built-in tool, and re-upload anything that arrived via CSV so it gets variants.
- Reduce font origins. Fewer third-party font domains mean fewer connections before text paints.
- Re-measure in the field, not the lab. Use Chrome field data or real-user monitoring and judge at the 75th percentile.
Steps one to three are where I would expect most of the gain. That is a recommendation from the metric weights above, not a measured result on your site.
Does Bandwidth Matter for Cost?
Yes, because bandwidth is part of the plan. Webflow's pricing page lists Premium bandwidth tiers from 50 GB up to 2.5 TB per month, against 10 GB on Basic and 1 GB on Starter. A heavy hero served to every mobile visitor now affects your plan tier as well as your score, which is one of the few places where performance advice and pricing point the same way.
When Is Webflow the Wrong Answer?
If your marketing site needs per-request personalization, a heavy interactive product demo or components shared with your app, you are fighting the tool instead of tuning it. WordPress versus a custom website lays out that fork, and the best CMS platforms for marketing sites shows where Webflow wins. For many small business sites the sensible move is to keep the visual builder and push logic elsewhere: pairing Webflow with Zapier runs automations off the page, so they add nothing to page weight.
If you are still costing the project, what a website build costs sets expectations, and practical advice for a successful website build covers process decisions that cause much of the damage. Performance is cheaper when it is designed in than when it is retrofitted.
Open your site in DevTools now and find the LCP marker. Is it the element you expected?