60 Percent of Webflow Sites Still Fail INP. Here Is the Fix Stack.

On May 7, May 17, and through this month, three independent audits from Lumien, Publive, and Digital Applied each documented that roughly six of ten sites still fail INP's 200 millisecond threshold. That is the two-year mark of Google making INP a Core Web Vital in place of First Input Delay on March 12, 2024. The data has not budged the way the industry expected.

I covered the positioning argument for treating INP as the primary performance metric in my INP primary-metric piece. This post is the diagnostic and tactical follow-up. Five fixes, in order of impact, that reliably drop a Webflow site below 200 milliseconds.

What Is INP and Why Did It Replace FID?

Interaction to Next Paint measures the latency between a user interaction (click, tap, keypress) and the next visual update on screen. It replaced First Input Delay because FID only measured the first interaction in a session, while INP captures every interaction and reports the worst one or the 98th percentile, whichever is more representative of the session.

Google made INP a Core Web Vital on March 12, 2024, which means it now factors into Page Experience signals and indirectly into search rankings. The good threshold is under 200 milliseconds. Needs improvement is 200 to 500. Poor is over 500. Most B2B SaaS marketing sites land in needs-improvement territory.

What Is a Good INP Score for a B2B SaaS Site?

Under 200 milliseconds is the Google "good" threshold and what every Phoenix Studio retainer site targets. Realistic best-in-class B2B SaaS sites land between 80 and 150 milliseconds on mobile. Anything above 300 milliseconds will feel sluggish to a user even if Lighthouse marks it as acceptable. The CrUX field data is what actually matters, not lab numbers.

The lab-versus-field gap is large. Lighthouse may report 120 millisecond INP in a local run while CrUX shows 380 milliseconds for the same site under real-world load. Always trust the CrUX numbers in Search Console over the Lighthouse score. Field data wins every argument with a stakeholder.

How Many Sites Still Fail INP Two Years Into the Rollout?

According to a Lumien analysis published May 7 and updated May 17, 2026, six of ten audited sites have INP above 200 milliseconds. Digital Applied citing CrUX data this month finds 43 percent of sites still fail the threshold. Publive's May 5 audit of publisher sites with analytics and ad scripts found only 37 percent achieve good INP on mobile.

The pattern is consistent. INP is the hardest Core Web Vital to pass on a content-heavy marketing site because every third-party script, every analytics tag, and every chat widget adds processing time to the interaction. Webflow sites are not exempt from this, especially when HubSpot, Intercom, and Cloudflare are layered on top.

How Does Webflow's GSAP Migration Affect INP?

Webflow announced native GSAP support in early 2026, which lets sites move animations off the legacy IX2 interaction system onto GSAP's ScrollTrigger and Timeline APIs. GSAP's rendering loop is more efficient than IX2 for complex animations, and the migration typically drops INP by 30 to 80 milliseconds on animation-heavy pages.

That is a meaningful gain on its own. The catch is that GSAP scripts add bundle weight, so the net effect depends on which animations you migrate. For Phoenix Studio retainers, the migration sequence I follow is to keep simple fades on IX2, move scroll-triggered sequences to GSAP, and remove anything that no longer serves a purpose. The audit pattern is the same as I described in my Core Web Vitals INP piece.

What Are the Five Fixes That Reliably Drop INP Below 200 Milliseconds?

One, audit third-party scripts and remove what is not essential. Two, defer non-critical scripts using the async or defer attribute. Three, split long tasks using scheduler.yield or requestIdleCallback. Four, lazy-load below-the-fold images with fetchpriority and modern formats. Five, migrate animation-heavy interactions from IX2 to GSAP where it makes sense.

The order matters because the first fix has the largest impact. Third-party scripts are responsible for 60 to 80 percent of INP regressions on Webflow B2B SaaS sites in my audit data. Cut HubSpot, Intercom, and a tag manager you no longer use, and INP drops measurably without touching any code on the site.

Why Is Lighthouse INP Misleading Versus CrUX Field Data?

Lighthouse runs a simulated interaction on a clean browser instance against your localhost or the deployed URL. CrUX collects real user data from Chrome users with their actual extensions, network conditions, and hardware. The two diverge most on sites with heavy third-party scripts because Lighthouse loads them in a clean environment while CrUX captures the cumulative effect on real devices.

The practical implication is to use Lighthouse for diagnosis and CrUX for measurement. Lighthouse will tell you which interaction is slow. CrUX will tell you how badly it affects your actual users. If your Lighthouse INP is green and your Search Console INP is red, trust Search Console.

How Do HubSpot Forms and Chat Widgets Damage INP?

HubSpot's forms script and Intercom's chat widget each add roughly 200 to 400 kilobytes of JavaScript that runs on page load and intercepts user interactions for tracking. Both scripts execute on the main thread, which means every click handler queues behind their work. INP scores on pages with both scripts loaded routinely cross 400 milliseconds.

The fix is to lazy-load both. HubSpot forms can be embedded as iframes with the loading="lazy" attribute. Intercom can be initialized on first user interaction rather than on page load. Both moves drop INP by 80 to 150 milliseconds on a typical SaaS landing page. The trade-off is a small delay before chat is available, which most users do not notice.

What Should You Do If Your INP Regressed After March 2024?

March 2024 is when INP became a Core Web Vital, but it is also when many Webflow sites added GA4 enhanced measurement and third-party heatmap tools that affect INP without being obvious in network panels. The forensic move is to compare your current third-party script inventory against an archive.org snapshot from February 2024 and identify what was added.

Once you have the diff, remove what was added with limited ROI. Heatmap tools like Hotjar and Smartlook are the most common culprit. They run continuous DOM observation in the background, which interferes with INP. If you are not actively using the recordings, remove the script.

How Do You Find INP Issues Inside Webflow Specifically?

Open Chrome DevTools, switch to the Performance tab, click the Record button, perform the interaction that feels slow, and stop recording. Look for long tasks in the main thread timeline. Webflow Interactions older than the May 2026 GSAP migration often show up as 80 to 200 millisecond blocks. Custom Code Embed blocks show up tagged with their script source.

For systematic auditing, the Chrome 149 Long Animation Frame API (covered in my Long Animation Frame for Webflow INP piece) gives a precise readout of which interactions are slow and which scripts caused them. That is the right tool for a Phoenix Studio audit handoff to a client.

When Should You Migrate Off Webflow for Performance Reasons?

Almost never. Webflow renders clean semantic HTML server-side and produces INP scores well under 200 milliseconds when third-party scripts are managed properly. The cases where migration makes sense are sites that need server-side rendering of personalized content for logged-in users, more than 10,000 CMS items in a single Collection, or real-time data updates.

For 95 percent of B2B SaaS marketing sites, the right move is to fix INP on Webflow rather than migrate off. The five-fix stack above resolves it in a half-day of audit and implementation, and the result is a faster site without giving up the editorial workflow Webflow provides.

If you want a Phoenix Studio audit of your specific INP score and the third-party script inventory dragging it down, drop me a line. Let's chat.