Every week, someone asks me if their slow site can actually be made fast. The answer is almost always yes, and it usually does not take as long as they think.
Most Webflow and CMS-based sites scoring in the 40-60 range on PageSpeed Insights can move to 80+ in a single focused day of optimization work. Not a redesign. Not a rebuild. Not a platform migration. A day of targeted fixes applied in the right order.
The reason is that most slow sites are slow for the same reasons. And those reasons have known, fast fixes.
What a realistic optimization day looks like
Here is the actual sequence I follow when a client sends me a site scoring 45 and asks me to fix it:
Morning: images (2-3 hours)
Images are the biggest win on nearly every site. I download every above-the-fold image, resize to display dimensions, compress to WebP at 75-80% quality, and re-upload. Then I set the hero image to eager loading and everything else to lazy.
This step alone typically accounts for 15-25 points of the total improvement. I wrote about the specifics in my post on image optimization for Webflow.
Midday: video and fonts (1-2 hours)
If there is an autoplay background video, I replace it with a poster image and defer the video load. I have written about the autoplay video trade-off in detail.
For fonts, I audit which weights are loaded versus used, remove the unused ones, set font-display to swap, and preload the critical weight. More on this in my post on fonts slowing your site.
Afternoon: scripts and verification (2-3 hours)
I audit every third-party script. Anything dead gets removed. Anything non-critical gets deferred. If there is a GTM container with 20+ tags, I review each one. I wrote about this specific problem in my post on marketing tags killing page speed.
Then I run PageSpeed Insights again, compare the before and after, and verify that field data metrics are on track. The lab score shows the immediate improvement. Field data takes 28 days to fully reflect the changes.
Where the 40 points actually come from
On a typical site moving from 45 to 85:
- Images: 15-25 points. Oversized images are almost always the single biggest contributor to slow LCP.
- Video: 5-10 points. Removing or deferring autoplay video directly improves LCP.
- Fonts: 3-5 points. Fewer font files mean faster rendering and less CLS.
- Scripts: 5-10 points. Trimming and deferring JavaScript improves both LCP (by freeing bandwidth) and INP (by freeing the main thread).
The total is not perfectly additive because the metrics interact with each other. But the ballpark holds: four categories of fixes, 30-40 points of improvement.
Want a website that turns visitors into customers, not just compliments?
Book a 15-min introWhy 100/100 is rarely worth chasing
I get asked about this constantly. "Can you get me to 100?" Yes, usually. But the last 10-15 points require trade-offs that are not worth making:
- Removing custom fonts entirely and using system fonts
- Eliminating all third-party scripts, including analytics
- Removing all animations and transitions
- Serving minimal CSS with no framework overhead
- Dropping all but the most basic images
A 100-scoring site often looks and feels stripped down. It sacrifices brand identity, analytics capability, and user experience polish for a number that does not directly correlate with better business outcomes.
The meaningful thresholds are the Core Web Vitals targets: LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1. If your field data passes those, you have "good" performance in Google's eyes. Whether your lab score is 85 or 100 does not change that assessment.
What this looks like in business terms
Here is the framing that matters more than the score:
A site scoring 45 with an LCP of 5 seconds has a bounce rate problem. Visitors are leaving before the page finishes loading. On a mobile connection, some visitors never see the hero section at all.
That same site scoring 85 with an LCP of 1.8 seconds loads before the visitor can decide to leave. The CTA is visible in under 2 seconds. Form submissions, clicks, and demo bookings all benefit because the page is usable faster.
The business impact is not "my PageSpeed score went up." It is "fewer people bounced, more people saw the CTA, and more people took action." That is what speed buys you.
When a day is not enough
Some sites have deeper issues that require more than a day:
Framework-heavy sites (React, Next.js, Angular) where the JavaScript bundle is 500 KB+ and INP is over 500ms. These need code-splitting, dependency auditing, and possibly architectural changes. I covered this in my post on making heavy framework sites fast.
Sites with hundreds of pages where the speed issues are template-level. Fixing one page template might fix 200 pages, but identifying and testing each template takes time.
Sites with deep CMS integration issues where the platform itself injects slow code that cannot be easily removed. Shopify's checkout scripts, WordPress plugin bloat, and Squarespace's rendering approach each have platform-specific constraints.
For these cases, a focused day gets you the quick wins (images, fonts, obvious scripts), and a longer engagement handles the structural issues. But even the quick wins are worth doing first because they often account for 60-70% of the total improvement.
Setting the right expectations
When a client asks "how fast can you make my site," here is what I tell them:
- Day one: 30-40 point improvement from asset optimization. Immediate lab score improvement, field data improvement over the next 28 days.
- Week one: additional 5-10 points from deeper script auditing and template-level fixes.
- Month one: field data validates the changes and shows real-world improvements in Core Web Vitals.
The fastest way to see if your site has these quick wins waiting is to run it through PageSpeed Insights and see if images, video, or JavaScript are the top recommendations. If they are, a focused day can transform the numbers.
If your site is stuck in the 30s or 40s and you want a realistic picture of what is achievable, that is exactly the kind of performance assessment I run. Honest expectations, targeted fixes, measurable results.