Agency advice on fonts tends toward the unfalsifiable. Some recommend Google Fonts, others insist on self-hosting everything, but nobody had measured what firms are actually doing in practice. We wanted hard numbers, not opinions.
In a single point-in-time crawl we scanned 303 law firm websites and analyzed their font loading strategies, file counts, preloading practices, and font-display usage. This is what we found.
Methodology
We collected font loading data from 303 law firm websites using automated browser analysis. For each site, our script examined:
- Google Fonts usage: whether the site loads fonts from fonts.googleapis.com or fonts.gstatic.com.
- Self-hosted fonts: whether the site serves font files from its own domain, detected via @font-face declarations and local .woff/.woff2/.ttf file references.
- Font file count: the total number of font files loaded during page render.
- Font families: the number of distinct typeface families in use.
- Preloading: whether any font files are preloaded using link rel="preload".
- font-display: whether any @font-face rules use the font-display property, particularly the "swap" value.
Sites were sorted into four groups: Google Fonts only, self-hosted only, both, or none detected. The "none detected" category includes sites using only system fonts, sites with fonts embedded in JavaScript bundles, and sites where our detection methods could not identify the font source.
The numbers at a glance
Font loading strategy breakdown
How a site loads its fonts determines its latency, reliability, and cache behavior. We placed every site into one of four groups based on where its font files originate.
| Strategy | Sites | Share of total |
|---|---|---|
| None detected | 104 | 34.3% |
| Self-hosted only | 85 | 28.1% |
| Both (Google + self-hosted) | 73 | 24.1% |
| Google Fonts only | 41 | 13.5% |
The largest group (34.3 percent) had no detectable custom font loading. Many of these sites use system font stacks, which is the fastest possible approach since no font files download at all. Others may load fonts through JavaScript or CSS-in-JS frameworks our static analysis could not see.
Self-hosted fonts (28.1 percent) offer the best balance of visual branding and performance when implemented correctly. Loading fonts from your own domain eliminates the extra DNS lookup and TLS handshake that Google Fonts requires, and gives you full control over caching, subsetting, and format selection.
The 24.1 percent of sites using both Google Fonts and self-hosted fonts are in the worst position from a performance standpoint. They pay the latency cost of third-party connections while also serving additional files from their own server. In most cases this happens because a WordPress theme bundles its own fonts while the site owner separately loads Google Fonts through a plugin or theme customizer.
Google Fonts usage patterns
Of the 303 sites analyzed, 114 (37.6 percent) use Google Fonts in some form. Among those that do, the family choices cluster tightly around a small set of popular sans-serif typefaces.
| Font family | Sites | Share of all 303 sites |
|---|---|---|
| Open Sans | 20 | 6.6% |
| Montserrat | 16 | 5.3% |
| Roboto | 10 | 3.3% |
| Lato | 10 | 3.3% |
| Poppins | 9 | 3.0% |
| Playfair Display | 7 | 2.3% |
| Merriweather | 4 | 1.3% |
| Oswald | 3 | 1.0% |
| Mulish | 3 | 1.0% |
| Roboto Slab | 3 | 1.0% |
| Raleway | 3 | 1.0% |
The top five families are all sans-serif typefaces optimized for screen readability. Playfair Display is the lone serif in the top group, often reserved for headings to lend a traditional or authoritative feel. The average site uses just 1.2 font families, which suggests most firms stay with a single typeface, often across several weights, rather than mixing heading and body fonts.
One pattern is worth noting. Several of these fonts (Open Sans, Roboto, Lato) are legacy choices from the early 2010s. A modern alternative such as Inter, designed specifically for screen interfaces, appears on only 0.7 percent of sites. That suggests many law firm sites run outdated themes or have not revisited their typography in years.
Preloading and font-display adoption
Font preloading and font-display: swap are the two most impactful optimizations for web font performance. Preloading tells the browser to begin downloading font files immediately, before the CSS is parsed. The font-display: swap property keeps text visible in a fallback font while the custom font loads, preventing the flash of invisible text (FOIT) that plagues poorly configured sites.
Adoption of both is remarkably low across the legal industry.
| Optimization | Sites | Share of total |
|---|---|---|
| Both preload + swap | 23 | 7.6% |
| font-display: swap only | 52 | 17.2% |
| Preload only | 19 | 6.3% |
| Neither | 209 | 69.0% |
69.0 percent of law firm websites use neither font preloading nor font-display: swap. These sites are fully exposed to the browser's default font loading behavior, which usually means invisible text for up to 3 seconds on slow connections (Chrome, Firefox) or an immediate fallback with a disruptive swap (Safari).
Only 7.6 percent of sites implement both optimizations together. That is the recommended best practice and the gold standard for font loading performance. For context, Google's own PageSpeed Insights flags missing font-display: swap as a performance issue, yet over 75 percent of these law firm sites would fail that check.
Font file count distribution
Every font file a browser must download adds to page weight and competes for bandwidth during the critical rendering path. Best practice recommends limiting font files to 2 to 4 total. Here is how law firm websites actually stack up.
| Font files | Sites | Share of total |
|---|---|---|
| 0 (no font files) | 188 | 62.0% |
| 1 to 3 files | 31 | 10.2% |
| 4 to 6 files | 24 | 7.9% |
| 7 to 10 files | 13 | 4.3% |
| 11+ files | 47 | 15.5% |
The most common result is zero detected font files (62.0 percent), which aligns with the 34.3 percent using no custom fonts plus sites where fonts load through mechanisms our scanner could not detect, such as JavaScript-injected stylesheets or iframe-embedded content.
Among sites that do load font files, the distribution is bimodal. A small group loads a reasonable 1 to 3 files (10.2 percent), while a surprisingly large 15.5 percent load 11 or more. The worst offender in our dataset loaded 113 font files on a single page. Sites in the 11+ category typically load every available weight and style of multiple families, often because a WordPress theme or page builder enqueues fonts broadly rather than loading only what the page actually uses.
The average across all sites is 5.9 font files. Among sites carrying at least one font file (115 sites), the average jumps to 15.5 files, well above the recommended ceiling of 4.
What this means for law firms
The lesson runs the same direction as every measurement above. Preload the fonts you actually need. Set font-display: swap so text never goes invisible. Self-host subsetted woff2 files instead of leaning on a third party. Limit yourself to the two or three weights the page uses, not every weight a theme happens to bundle. None of this is exotic, yet 69.0 percent of the industry does none of it.
This is the gap between a site assembled from a theme and a site engineered to a standard. Every Nitrosite self-hosts subsetted, content-hashed woff2 fonts, preloads the critical files, and sets font-display so text appears instantly. That is part of how a Nitrosite reaches a 0.2 second first contentful paint and 0.00 cumulative layout shift while the average law firm site scores 34 on Lighthouse. And it is the kind of deficit our method surfaces before any design begins: NitroCMS unifies five data sources, Google Analytics, Search Console, Google Ads, independent SEO and competitor intelligence, and the Constellate Analytics Engine, our first-party analytics, and hands a technical audit of up to 500 pages to Claude Opus, which flags slow pages and render-blocking resources alongside everything else. Evidence over guesswork.