font-display on React sites

Here, a site counts under every font-display value present in its @font-face rules. Sites with at least one matching rule have a good LCP: swap at 79%, fallback at 73%, auto at 70%, block at 66%. Why does swap (79%) sit right next to auto (70%)? Three in four sites that load Google Fonts with swap also load icon or theme fonts with block or auto. Those multi-font pages count under both values.

Here, a site counts under every font-display value present in its @font-face rules. Sites with at least one matching rule have a good CLS: swap at 81%, block at 73%, auto at 73%, fallback at 71%. Why does swap (81%) sit right next to auto (73%)? Three in four sites that load Google Fonts with swap also load icon or theme fonts with block or auto. Those multi-font pages count under both values.

Across 9,533 React sites that declare web fonts, swap accounts for 71.3% of all @font-face rules, auto for 21.9%, block for 6.3%, fallback for 0.4%, and optional for 0.1%. At the site level, 32.9% of sites mix swap with blocking rules on the same page, 8.5% use only blocking rules (auto, block, or fallback), and 58.6% use swap or optional on every font rule.

When each site is counted once by its full rule mix, sites that keep every font on swap or optional have a good LCP on 90% of sites, against 79% for blocking-only sites and 69% for mixed sites. Mixed sites load a median of 6 font files across 59 @font-face rules (usually Google Fonts plus Font Awesome or theme icon fonts). Pure swap sites load a median of 5 font files across 39 rules.

When each site is counted once by its full rule mix, sites that keep every font on swap or optional have a good CLS on 89% of sites, against 73% for blocking-only sites and 73% for mixed sites. Mixed sites load a median of 6 font files across 59 @font-face rules (usually Google Fonts plus Font Awesome or theme icon fonts). Pure swap sites load a median of 5 font files across 39 rules.

Comparing sites that load the same number of font files separates font-display from font count. With 1 to 2 font files, sites on pure swap have a good LCP on 88% of sites, against 70% for sites with any block rule. On pages loading 8 or more font files, pure swap reaches 83% against 58% for sites with block.

Comparing sites that load the same number of font files separates font-display from font count. With 1 to 2 font files, sites on pure swap have a good CLS on 84% of sites, against 69% for sites with any block rule. On pages loading 8 or more font files, pure swap reaches 87% against 74% for sites with block. At every font count, pure swap matches or beats block on field CLS.

On Wix, 99% of declared @font-face rules use font-display: swap, 1% use auto, and 0% use block. 92% of Wix sites have a good LCP. On WordPress, 49% of rules use swap, 39% use auto, and 11% use block, and 66% of WordPress sites have a good LCP.

On Wix, 99% of declared @font-face rules use font-display: swap, 1% use auto, and 0% use block. 96% of Wix sites have a good CLS. On WordPress, 49% of rules use swap, 39% use auto, and 11% use block, and 75% of WordPress sites have a good CLS.

Platform@font-face rules on its sitesGood LCP

React sites with font-display 9,533 sitesswap 71%auto (browser default) 22%79%

Wix 5,822 sitesswap 99%92%

WordPress 2,333 sitesswap 49%auto (browser default) 39%block 11%66%

swap auto (browser default) block fallback optional

Does the platform explain the font-display gap?

On Wix, 95% of sites that keep every font on swap or optional have a good LCP, compared with 100% for blocking-only sites and 80% for Wix sites that mix swap with blocking rules. Within Wix, keeping every font on swap beats mixing rules by 15 percentage points. On WordPress, 70% of sites that keep every font on swap or optional have a good LCP, compared with 74% for blocking-only sites and 63% for WordPress sites that mix swap with blocking rules. Within WordPress, pure swap and mixed sites have a good LCP about equally often.

On Wix, 96% of sites that keep every font on swap or optional have a good CLS, compared with 100% for blocking-only sites and 94% for Wix sites that mix swap with blocking rules. Within Wix, pure swap and mixed sites have a good CLS about equally often. On WordPress, 72% of sites that keep every font on swap or optional have a good CLS, compared with 81% for blocking-only sites and 74% for WordPress sites that mix swap with blocking rules. Within WordPress, pure swap and mixed sites have a good CLS about equally often.

How font-display affects LCP and CLS during page load

When text is the LCP element, font-display: block and font-display: auto hide it for up to 3 seconds while the font file downloads. If the font arrives late, that wait delays LCP directly. With font-display: swap, the browser paints the fallback system font right away so LCP is recorded on first render, then swaps in the web font once it downloads. Swapping an unadjusted fallback font can shift text lines. Even so, pure swap sites match or beat blocking sites on field CLS at every font count (89% good CLS).