The latest articles on DEV Community by Infinite Semantic Inc. (@infinitesemantic-com). https://dev.to/infinitesemantic-com https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4097379%2F9dfb2c38-2af7-4280-9d9b-8f2280eaa5f2.png https://dev.to/infinitesemantic-com en Infinite Semantic Inc. Thu, 27 Aug 2026 12:29:23 +0000 https://dev.to/infinitesemantic-com/we-measured-139-small-business-websites-3-meet-googles-speed-threshold-2k48 https://dev.to/infinitesemantic-com/we-measured-139-small-business-websites-3-meet-googles-speed-threshold-2k48 <p>Every web developer tells clients their site is too slow. I got tired of saying it without a number behind it, so we measured.</p> <h2> The method </h2> <p>We took the top organic results for 12 local service queries across 12 US cities — plumbers, dentists, salons, that kind of business. We kept only the businesses' own sites: no directories, no aggregators, no national chains. Then we identified each site's platform from its page source and ran one Lighthouse mobile audit per site through the PageSpeed Insights API.</p> <p>139 sites measured, August 2026.</p> <h2> The results </h2> <p>The median site takes <strong>8.9 seconds</strong> to render its largest content. Google's threshold for "good" is 2.5 seconds.</p> <p>Four sites out of 139 met it. That's 3%.</p> <p>The median page ships <strong>1.2 MB of JavaScript</strong> before doing anything the visitor asked for. A quarter of the sample scored below 50 on Lighthouse performance. The median mobile score was 57.</p> <h2> What it isn't </h2> <p>It's lab data — one run per site, one point in time, US only. That's enough to describe a population, not precise enough to rank two individual sites against each other. Every caveat is listed in the write-up.</p> <h2> The part I didn't expect </h2> <p>Slow sites showed up on every platform. WordPress, the builders, sites we couldn't identify — the medians moved around but nobody was clean. That points at how sites get assembled rather than at any one vendor: a theme, a page builder, a chat widget, tracking scripts, a booking embed, sliders and fonts, each added by someone solving one problem without seeing the total.</p> <h2> The data is open </h2> <p>Every domain, its detected platform, the query it was found through and all measured values are published as a CSV. Re-analyse it, check my arithmetic, or run the same queries in your own city:</p> <p><a href="https://insemantic.com/data/small-business-website-speed-2026-08.csv" rel="noopener noreferrer">https://insemantic.com/data/small-business-website-speed-2026-08.csv</a></p> <p>Full write-up with the method and every caveat: <a href="https://insemantic.com/blog/small-business-website-speed-study/" rel="noopener noreferrer">https://insemantic.com/blog/small-business-website-speed-study/</a></p> <p>If you spot a hole in the methodology, I'd rather hear it than not.</p> webdev performance webperf showdev