Next.js detected on 76.3% of framework-identifiable homepages in 10,242-site crawl

A snapshot of one directory's stacks

directree Research scanned the public homepages of 10,242 software websites in September 2026 and detected Next.js on 3,863 of them, according to a study published on dev.to. That is 76.3% of the 5,060 sites where the scanner could identify at least one framework, and 37.7% of every site checked.

The headline figure comes with important framing. The scanner found no recognizable framework signal on just over half the sample: 5,060 of the 10,242 checked hostnames (49.4%) showed at least one framework signal, while 5,182 (50.6%) showed none. directree stresses that a missing signal only means it could not match a supported framework, not that the site uses none, since Next.js, for example, lets sites disable the response header that identifies it. A further 368 published listings had no recorded stack check and were excluded.

The ranking

Among sites where something was detected, Next.js led by a wide margin. React, matched through its own signature, appeared on 830 sites (16.4% of the detected group), followed by Laravel at 235 (4.6%), Vue at 198 (3.9%), Nuxt at 184 (3.6%), Django at 88 (1.7%), SvelteKit at 81 (1.6%), Ruby on Rails at 74 (1.5%) and Angular at 30 (0.6%).

These rows overlap, with 481 websites showing multiple framework signals, so the percentages cannot be summed into a market-share total. The React figure also deserves a footnote: Next.js is itself a React framework, and directree's stored data omits signals inferred purely from another detected technology, so the 830 counts only independently matched React signatures. There were 574 sites with a React signal and no Next.js signal, and the same distinction applies to Vue and Nuxt.

How the measurement worked

The observations come from published directree listings collected between 6 and 15 September 2026, with 8,944 of the 10,242 stack checks dating to 6 September. Each listing is a distinct hostname, spanning 10,178 registered domains, and the directory covers software broadly, from AI products to developer tools. The sample follows directory membership rather than a random draw from the web.

The scanner inspects public responses, such as headers, cookies and HTML, for recognizable clues including Next.js's response header and characteristic page data or asset paths. It does not run a full browser session or inspect code behind a login, so the results describe the pages visitors arrive at, not necessarily the underlying application's stack. directree also did not manually audit each site's source code, and notes that some signatures are more distinctive than others.

A sensitivity check grouping by registered domain instead of hostname produced a nearly identical result: 3,854 of 5,040 domains with a detected framework showed Next.js, or 76.5% versus the headline 76.3%.

Category differences

Among the 16 categories with at least 200 checked sites, the highest Next.js shares of all checked sites were in AI Image & Video Generation (61.1% of 488 checked; 77.8% of those with a detected framework), Video & Audio (52.6% of 384) and Directory & Listings (51.1% of 309). Within the detected subset, Design & UI had the highest share at 83.0%.

Developer Tools sat at the other end: Next.js on 93 of 302 checked sites, or 30.8%, and 66.9% of the detected subset. directree notes it did not ask those teams why they chose their stacks, and the category gaps could reflect website architecture, templates, the products listed, or how easily a framework can be detected.

Why it matters

The numbers are a concrete data point on how concentrated the React-and-Next.js toolchain has become among software product homepages, at least within this directory's population. For teams building integrations, templates or documentation, directree suggests the 3,863 listed hostnames offer a starting point for scoping a Next.js audience.

The caveats matter just as much. The 76.3% figure describes only the half of the sample with a detectable framework, the crawl is not a random sample of the web, and the study measured detection rather than performance, revenue or developer satisfaction, so it cannot tell anyone which framework to choose. directree recommends repeating the queries quarterly, and warns that a future percentage can only be read as an adoption change once directory membership and detection rules are accounted for.