Lighthouse 95 Score Without Removing Animations or Fonts

You can score Lighthouse 95+ and keep your animations and custom fonts. The score is almost never lost to design — it's lost to oversized images, render-blocking JavaScript and a slow server. Fix those three, load your fonts and motion the smart way (which we'll walk through below), and you get speed and polish together. The advice to "just strip the font and the animations" is lazy: it's the easiest thing to delete, not the thing that's actually slowing you down.

Here's the trap. Lighthouse hands you a single number, the number feels like a grade, and a grade demands action — so people start cutting the visible, brand-defining bits because those are easy to see. Meanwhile the real weight sits in files they never look at. Let's fix the right things.

The score is a means, not the goal

Lighthouse is genuinely useful — it gives you a snapshot and a prioritised list of fixes. But the number is a proxy for what actually pays you: a fast, usable page that converts. Speed matters because it moves money, and the biggest companies have proven the link to the millisecond. An ex-Amazon engineer, Greg Linden, found every 100ms of extra delay cost Amazon around 1% in sales. Google's Marissa Mayer reported a half-second delay caused a 20% drop in traffic. The BBC lost 10% of its users for every additional second of load time.

Notice what none of those companies did to fix it: delete their branding. They fixed the plumbing. A site can score 95 and still convert badly, and a site in the 60s can sell all day — but the goal isn't a low score either. The goal is a fast site that still looks like you. Both are achievable at once.

Why animations and fonts are usually worth keeping

Custom fonts and motion aren't the enemy. A consistent brand font signals you're a real, established business, not a template someone bashed out in an afternoon. A subtle animation on a call-to-action draws the eye to the one thing you want clicked. Picture an Ipswich bakery whose warm, rounded headline font is half the reason the brand feels family-run and welcoming — strip it for a generic system font and the page technically loads faster but feels like nobody owns it.

Fonts carry emotion: a serif reads as established and trustworthy, a clean sans-serif reads as modern. Motion, used sparingly, guides attention and makes a page feel responsive. Delete all of it to chase a number and you can win two Lighthouse points while losing the thing that made a visitor trust you enough to enquire. That's a bad trade — and the good news is you almost never have to make it.

The real bottlenecks (where the points actually are)

Most sites fail Lighthouse for reasons that have nothing to do with branding:

• Oversized images. A single un-compressed hero photo can weigh several megabytes — by far the most common offender. This is where the bulk of your lost points live.
• Render-blocking JavaScript. Scripts that load before the visible content holds the whole page hostage, and it's worst on mid-range phones over mobile data.
• Slow server response. If the server takes a second or two just to start sending the page, no front-end trick can save you. Fix this first.
• Too many requests. Every separate image, stylesheet and script is another round trip; dozens of them add up.
• Bloated, unused CSS. Page builders and themes ship styles for features you never use, and the browser parses all of it anyway.

Compress your images properly and you'll typically claw back more points than any other single change — usually with zero visible difference to the design. Fix the server and the scripts and you're most of the way to 95 before you've touched a font.

Does it work before it's perfect?

Before you chase the last few points, ask the only question that pays the bills: does the site work? Does it load in a reasonable time, can people find what they want, and do they enquire? If yes, an extra Lighthouse point won't change your week. Plenty of sites in the 60s generate leads daily; plenty in the 90s convert nobody. Get the fundamentals — images, server, scripts — solid first, then use fonts and animation deliberately rather than fearfully.

How to keep fonts and animations and still score 95

You get both with a more careful build, not by deleting things:

1. Use font-display: swap; so the page shows readable fallback text immediately and swaps in your brand font when it's ready. No blank wait, no penalty.
2. Serve modern font formats (WOFF2). It compresses far better than older TTF/OTF files, so your brand font downloads quickly.
3. Subset your fonts. Ship only the characters you use. An English-only site doesn't need Cyrillic or Greek glyphs padding out the file.
4. Lazy-load below-the-fold animations. Don't spin up motion the visitor can't see yet.
5. Prefer CSS animations over heavy JavaScript ones. They lean on the browser's own rendering engine and stay smooth.
6. Preload your hero font. A single <link rel="preload"> tells the browser to fetch it early so headings don't flash.

When stripping them out IS the right call

Sometimes the honest answer is yes, cut it. Make that call deliberately, not by reflex:

• A genuinely slow server. If response time is consistently a couple of seconds, fonts and motion are noise — fix hosting first.
• Accessibility. Some motion triggers discomfort or seizures for users with vestibular conditions. Respect prefers-reduced-motion and drop any animation that can't be made safe. This is non-negotiable.
• A huge, unsubsettable font. If a display font is a couple of megabytes and won't shrink, find a lighter alternative — beauty isn't worth a multi-second wait.
• Decorative motion that earns nothing. If an animation doesn't guide, delight or convert, it's just cost. Cut it.

How a font experiment should actually run

Picture a Springwood accounting firm worried its brand font is hurting it. The right move isn't to guess — it's to A/B test. Keep the brand font but add font-display: swap; and a subset version on one variant, run a system-font version on the other, and split your traffic for a couple of weeks. Then read what real visitors did, not what a tool predicted. Nine times out of ten the optimised brand-font version wins on enquiries while still scoring well — which is exactly the point: you rarely have to choose. Measure with your own visitors before you delete anything.

The metrics Google actually ranks on

Google ranks on Core Web Vitals — Largest Contentful Paint (how fast your main content appears), Interaction to Next Paint (how quickly the page responds when tapped), and Cumulative Layout Shift (how much the page jumps around as it loads). They've been an official ranking factor since June 2021. (INP replaced the older First Input Delay metric in 2024, so ignore any guide still telling you to optimise for FID.) These are closer to real experience than the headline Lighthouse number, and LCP is usually where image and server work pays off most. Notably, a slow font swap is a classic cause of layout shift — so loading fonts the right way improves CLS and keeps your branding. Speed and polish, same fix.

Common mistakes when chasing the score

• Cutting branding first. It's the easiest thing to delete and almost never the real problem.
• Ignoring the server. Front-end tweaks can't outrun slow hosting.
• Sacrificing accessibility for a number. A high score that locks people out is a failure, not a win.
• Following every recommendation blindly. Lighthouse lists are prioritised for you — fix the big-impact items at the top and don't sweat the trivial ones.

If you're deciding right now

Be clear about what you actually want: not a number, but a fast site that still looks and feels like your business. Get the fundamentals solid — compress images, sort the server, defer non-critical scripts — then load your fonts and animations the careful way above. You almost never have to choose between speed and polish. Anyone telling you to gut the design to win two points is fixing the wrong thing.

You almost never have to choose between speed and your brand. The score is lost in the plumbing, not the paint.

If you want a straight answer on what's really costing your site its points, that's exactly what our free audit is for — and our pricing is published up front, with no lock-in and the code yours to keep.