The figures below are from W3Techs on 20 October 2026, and from the HTTP Archive 2025 Web Almanac (July 2025 crawl, published January 2026).
What the surveys count
W3Techs reports two percentages. All sites includes everyone in the survey. CMS share is the share among sites where a monitored CMS was detected. On 20 October 2026, 32.1% of sites had no monitored CMS. They sit in the all-sites figure, outside CMS share.
The Web Almanac uses the HTTP Archive crawl and Wappalyzer fingerprints. Its CMS breakdown shows a detectable CMS on 55% of desktop sites and 54% of mobile sites in 2025. Headless fronts and many static generators are placed in "none".
Content Management Systems in 2026
| System | Language | All sites | CMS share |
|---|---|---|---|
| No monitored CMS | n/a | 32.1% | n/a |
| WordPress | PHP | 39.6% | 58.3% |
| Shopify | Ruby | 5.3% | 7.9% |
| Wix | Scala | 4.2% | 6.2% |
| Squarespace | Java | 2.4% | 3.6% |
| Joomla | PHP | 1.1% | 1.6% |
| Webflow | Node | 0.8% | 1.2% |
| Duda | Java | 0.7% | 1.1% |
| Drupal | PHP | 0.6% | 0.9% |
| PrestaShop | PHP | 0.5% | 0.7% |
| OpenCart | PHP | 0.3% | 0.5% |
| Lovable | TypeScript | 0.2% | 0.4% |
| Astro | JavaScript | 0.2% | 0.3% |
| Framer | JavaScript | 0.2% | 0.3% |
Share of all sites, and of sites where a monitored CMS was detected. W3Techs, 20 October 2026.
Read the first number as "of the whole web" and the second as "of sites with a CMS the survey can name." WordPress is still leading in 2026 at 39.6% of all sites and 58.3% of CMS share. Lovable, Astro, and Framer are on the list because they are now large enough to count.
| System | 1 Jan 2015 | 1 Jan 2022 | 20 Oct 2026 |
|---|---|---|---|
| WordPress | 60.7% | up65.2% | down58.3% |
| Shopify | 0.7% | up6.6% | up7.9% |
| Wix | 0.3% | up2.9% | up6.2% |
| Squarespace | 0.5% | up2.7% | up3.6% |
| Joomla | 8.6% | down2.6% | down1.6% |
| Drupal | 5.1% | down2.0% | down0.9% |
CMS share in 2015, 2022, and 2026, from the yearly W3Techs series. Arrows compare each date with the one before.
The last decade shows the shift from self-hosted CMS to hosted builders like Shopify, Wix, and Squarespace. As platforms matured, the cost of a self-hosted solution became less attractive. Custom theme builders and complex integrations made the choice of a monthly subscription easier.
The last twelve months are worth separating. Across the decade, hosted builders and Shopify took share from the general open-source CMS. Across the last year, most of the drop in WordPress lined up with growth in sites that have no named CMS. What we see is a shift to AI-assisted scaffolding and self builds.
The shift to AI-assisted scaffolding and self builds is a noticeable trend. You've seen it on every feed. Even excluding Lovable, Astro, and Framer, the timeline is filled with vibe-coded startups and established brands building their own sites. It's a shift towards your agent or harness being your CMS.
Ecommerce Systems in 2026
| Store system | 1 Oct 2025 | 20 Oct 2026 |
|---|---|---|
| WooCommerce | 51.1% | down47.4% |
| Shopify | 27.9% | up32.3% |
| PrestaShop | 3.4% | down2.8% |
| OpenCart | 2.3% | down1.9% |
Share of detected stores from 1 October 2025 to 20 October 2026. Arrows compare the year.
WooCommerce is still the largest store platform, with Shopify taking ground year on year from the self-hosted carts. The all-sites ecommerce series and the 2025 Web Almanac ecommerce chapter rank the same four in the same order.
Using a provider to manage your ecommerce will likely be the popular choice for many years to come. Especially as AI-assisted purchase models mature, such as x402, providers are more set to offer agentic workflows, the necessary edge AI security, and the growing infrastructure agents and harnesses need.
What the top 10,000 sites use in 2026
Overall share does not show where the most traffic is. On 20 October 2026, of sites with a known CMS (WordPress, Drupal, Adobe, and WooCommerce, Shopify, Drupal), the top sites' CMS landscape looks like this:
- WordPress is 58.3% overall, 48.3% of the top million, and 50.5% of the top 10,000.
- Adobe Systems is 0.9% overall and 6.9% of the top 10,000.
- Drupal is 0.9% of known CMS sites overall and 6.7% of the top 10,000.
- Shopify is 7.9% overall, 15.8% of the top million, and 4.1% of the top 10,000.
- WooCommerce is 11.5% overall and 1.3% of the top 10,000.
Expect the top 10,000 to stay on a different list to the open web for a while. Drupal and Adobe keep a share several times their place on the open web, and these brands are not going to change their CMS quickly. Those who adopt AI agents as their CMS will pave the way for others, showing how this can be done at scale. When the output is brand content in public view, the stakes are high and systems will need to be ready for this. Read more about proofs and AI in Provable content vs AI slop.
What this can mean
Commerce - The shop is leaving the self-hosted cart. Shopify's increase, beside the decline in OpenCart, PrestaShop, and WooCommerce's share of stores, is the clearest move in the data. Checkout, catalogue, and apps now live mostly in a hosted admin. WooCommerce remains the largest store system, but with a declining trend.
Hosting - Wix and Squarespace spent a decade turning "I do not want to run a server" into CMS share. Webflow has become the studio version of this, and has been flat in the 2026 W3Techs series. Duda is the white-label version, still a small rise, and often sold by agencies as many sites under one account.
Custom builds - This is the open question, and the one most tied to the last year. Sites with no monitored CMS rose by 3.5 points while WordPress fell by 3.7 points of all sites. This is the work that never announces a CMS: static generators, headless fronts, and one reading of this data is a noticeable increase in AI-assisted scaffolding and self builds. That's excluding Lovable, Astro, and Framer, which are on the list because they are now large enough to count, but they are still a fraction of one percent.
Another reading is detection. On the longer 2021-2025 window, the Web Almanac shows CMS adoption rising, which means sites without a detected CMS were shrinking across those five years. A site becomes "a CMS site" the day the survey learns to see it. Hold the self-build story as plausible. Hold it lightly as the main cause. The decade in the yearly table is hosted builders and Shopify taking share from Joomla, Drupal, and, since 2022, from WordPress.
The top of the web - Drupal and Adobe are several times more common in the top 10,000 than on the open web. A university, a public body, and a national brand are choosing from a different list than a new brochure site. Overall decline for Drupal is largely the long tail leaving, but the busy sites remain.
Managed defaults - The Web Almanac's mobile Core Web Vitals comparison for 2025 puts about 45% of WordPress sites in the good band. Wix moved from about 55% to 74%. Duda sat near 85%. A platform that controls the theme, the host, and the scripts can lift the median. An open CMS leaves that work with the people who run the site. Some of the share moving towards builders is that operational saving, paid as a subscription.
The price of a first version - Scaffolding a site is cheap. Owning the architecture remains expensive: hosting, updates, accessibility, backups, and an editor and business logic a colleague can repeat next year. A rise in self-build activity is a sign that AI-assisted scaffolding is now good enough to deliver the first version. And it may not be long before we see a generation of systems that have no declared dependency at all.
What to do with a site you already run
Keep the system when it still covers the requirements, and ensure you are on top of your digital security (post coming soon on what that means for you and your business). That is the honest advice.
Check that your system is on a release line that still receives security updates. This is the most important action to take right now. Our recent post Zero-day PHP vulnerabilities in the wild covered two critical flaws, and gives you a clear reason you need to stay on top of your DevOps! An old version is a vulnerability: core, plugins, and modules that no longer get patches.
A self-build is a great option when you want to own the release and iteration lines, and you can carry the extra work to replace your existing CMS. Using an AI assistant you can scaffold a site quickly, then continue to use your assistant to maintain it without needing or exposing a backend. But today this is not reliable or scalable, and it needs strong safeguards and discipline. Someone still needs to patch it, host it, keep it accessible, and guide the business logic unless you're comfortable leading the fringe in AI-assisted workflows.
Editors and agents going headless
The admin, editors, and content-provider workflows around the system are the part that is changing the most, and a place where you can save a huge amount of time right now. Keeping your existing CMS layer and utilising AI agents to manage your content and submit to your own system APIs can boost your productivity immensely. The traditional web login will fade away as agents are given access to your CMS, making them a headless system for your content needs.
Put a small MCP server in front of your API, with the actions an editor already has: list, draft, update, submit. Register it once in the client. The same server works in Cursor, Claude, Claude Code, VS Code with GitHub Copilot, ChatGPT, Codex, Gemini CLI, and Antigravity CLI. Many agents, one door to your content managmenet layer.
Arc is the version we are building; connecting your CMS and tools to agents that can reliably manage your content. Agents for everyday business is the same idea at the width of a company, and the landscape, not a score is the picture those agents should be reading to make reliable decisions.