Best CMS for SEO in 2026: 12 Vendor Blogs Measured

Every article ranking for "best cms for seo" is a list, and six of the ten results are lists written by a company that sells one of the entries. Agility CMS ranks third with a list where Agility is first. Storyblok ranks ninth with a list where Storyblok is first. Neither list is dishonest about who wrote it. Both are useless for deciding anything.
So this article does something the lists cannot. Every CMS vendor runs its own blog on its own product. That blog is the best demonstration the vendor will ever give you, because they have the expertise, the incentive and root access. On 2026-11-18 we fetched one blog post from twelve of them and recorded what came back.
Two of the twelve emit no structured data at all. One has no meta description. One ships a <head> that is 1.2 MB long. The platform that every SEO list calls the worst choice emits the most complete markup in the sample.
The question "best CMS for SEO" is asking the wrong thing
A CMS cannot rank. It can only prevent you from doing the things that make ranking possible, or get out of your way. That distinction is the whole answer, and the data in the SERP supports it.
Rankability published the only real study in these results on 2026-10-17. They pulled top-ranking domains for 102 high-CPC commercial keywords, fetched 631 of them, and classified 524 with their own fingerprint detector. The largest single group was not a CMS:
| Platform | Share of top-ranking domains | Domains |
|---|---|---|
| No CMS detected | 28.4% | 149 |
| WordPress | 20.0% | 105 |
| Next.js | 13.7% | 72 |
| Adobe Experience Manager | 10.3% | 54 |
| Shopify | 6.7% | 35 |
| Drupal | 5.7% | 30 |
| Webflow | 1.7% | 9 |
| Wix | 0.4% | 2 |
| Craft CMS | 0.4% | 2 |
Their AI-citation sample points the same way. Across 167,272 cited AI answers they classified 209 sites, and 44.0% of the sites AI assistants cite run no detectable off-the-shelf CMS. Rankability says so themselves in their limitations section: "CMS choice is not itself a ranking factor."
Google has never named a preferred CMS either. Its SEO Starter Guide mentions a CMS twice, both times to say the CMS might save you work ("If you use a CMS, you might not need to do anything technical to your titles"), and never to recommend one. The same page is blunt about what platform choice buys you: "There are no secrets here that'll automatically rank your site first in Google (sorry!)."
That is the reframe. Stop asking which CMS is best. Ask whether yours gives you eight specific controls, and go measure.
The eight controls that actually matter
Each of these is a thing a CMS either hands you or withholds. Each is checkable from a terminal in under a minute, on your own site, today.
| Control | Why it decides anything | How to check |
|---|---|---|
| URL path | The only choice that is expensive to reverse | Look at a live post URL |
| Title and meta description per page | What the searcher reads before clicking | curl the page, grep the head |
| Canonical tag | Which of your duplicate URLs Google keeps | grep rel="canonical" |
| 301 or 308 redirects you control | Survival of any URL change | curl -I an old URL |
Sitemap with accurate lastmod |
How fast changes get re-crawled | Fetch /sitemap.xml |
| Structured data on the page type | Whether the page is machine-legible | Count ld+json blocks |
| Server-rendered body | Whether crawlers see your text at all | Count words in raw HTML |
| Content export | Whether you can ever leave | Try the export |
Nothing on that list is exotic. What is striking is how many platform vendors fail their own checklist on their own blog.
What twelve vendor blog posts actually emit
One post per platform, fetched 2026-11-18 with a desktop user agent, no JavaScript executed. "Raw words" counts the visible text present in the HTML the server sent, which is the text a crawler gets without rendering.
| Platform (page measured) | HTML bytes | <script> |
Raw words | Meta desc | Canonical | ld+json blocks |
Article or BlogPosting |
|---|---|---|---|---|---|---|---|
| WordPress 7.2-RC1 (wordpress.org/news) | 167,164 | 26 | 746 | none | yes | 0 | no |
| HubSpot CMS (blog.hubspot.com) | 341,133 | 41 | 4,214 | 202 chars | yes | 2 | yes |
| Craft CMS (craftcms.com/blog) | 134,402 | 7 | 1,993 | 141 | yes | 1 | no |
| Sanity (sanity.io/blog) | 136,945 | 22 | 1,332 | 75 | yes | 0 | no |
| Shopify (shopify.com/blog) | 584,221 | 8 | 2,013 | 125 | yes | 1 | no |
| Wix (wix.com/blog) | 2,621,912 | 82 | 7,939 | 152 | yes | 6 | yes |
| Webflow (webflow.com/blog) | 397,713 | 51 | 1,338 | 124 | yes | 2 | yes |
| Squarespace (squarespace.com/blog) | 406,167 | 48 | 3,005 | 161 | yes | 6 | yes |
| Storyblok (the #9 result) | 537,944 | 15 | 3,366 | 156 | yes | 2 | yes |
| Agility CMS (the #3 result) | 314,897 | 44 | 4,114 | 149 | yes | 3 | yes |
| Astro, on Vercel (the #4 result) | 120,218 | 18 | 2,585 | 146 | yes | 2 | yes |
| WordPress (the #8 result) | 92,586 | 25 | 1,763 | 147 | yes | 1 | no |
Four readings from that table, in order of how much they contradict the lists.
WordPress.org's own news post has no meta description and zero structured data. The <meta name="generator"> reads WordPress 7.2-RC1-65564, so this is the newest WordPress there is, run by the people who write it. There is an og:description and a correct canonical, and that is all. WordPress is the most common real CMS among top-ranking domains, and its own publication ships none of the markup every WordPress SEO plugin exists to add. The platform's ceiling and the platform's defaults are not the same thing, which is the entire argument for doing your own SEO rather than trusting a plugin.
Wix emits the most complete markup in the sample. Six JSON-LD blocks carrying BlogPosting, FAQPage, HowTo, HowToStep, VideoObject and SeekToAction, a 152-character meta description, a correct canonical. It also ships 2.6 MB of HTML and 82 script tags, and its </head> does not arrive until byte 1,212,831. The head alone is bigger than nine of the other eleven entire pages. Both facts are true at once, which is why a platform verdict built on one number is worthless. We looked at Wix on its own terms separately.
Shopify's blog post marks itself up as a PodcastEpisode and nothing else. No Article, no BlogPosting. The post is an interview, so PodcastEpisode is defensible, but it means the page offers Google no article semantics at all. Shopify ships plenty of commerce schema on product pages; the blog is a different template and nobody went back to it.
Craft CMS ships 7 script tags, the leanest page in the set, and no article markup. Its single JSON-LD block declares ProfessionalService and WebApplication, generated by the SEOmatic plugin, on a page that is a blog post. A plugin was installed, configured once, and never matched to the page type.
Two platforms could not be measured. drupal.org/blog returns a Fastly "Client Challenge" page that requires JavaScript, and contentful.com/blog returns a Vercel Security Checkpoint. Both are bot protection, not CMS behaviour, and we are not going to guess what sits behind them.
The result that should end the "which CMS ranks" question
The fourth result in this SERP, with the only original research on the page, runs on Astro deployed to Vercel, which is not a CMS. The ninth result, published by a headless CMS company, is served by Netlify. A headless CMS does not render pages. It hands JSON to a front end, and the front end decides everything in the eight-control table above. If you choose Contentful, Sanity or Storyblok, you have not chosen how your pages behave. You have postponed the choice.
URL shape is the one decision you cannot cheaply undo
Everything else in the table can be fixed with a template edit this afternoon. URLs cannot, because changing one costs you the redirect, the re-crawl and an interval of ranking you do not control. Google's own guidance on moving URLs is explicit: use "HTTP permanent redirects if possible, such as 301 and 308", keep them "generally at least 1 year", and "Expect temporary fluctuation in site ranking during the move."
Hosted platforms spend that budget for you at signup, by inserting a path segment you cannot remove. Measured on live sites, not read from documentation:
| Platform | Blog post URL shape | Evidence |
|---|---|---|
| Wix | /post/<slug> |
20 posts on livxmedia.com |
| Shopify | /blogs/<blog-handle>/<article> |
allbirds.com sitemap: /blogs/the-perch, /blogs/news |
| WordPress (default on wordpress.org) | /news/2026/11/<slug>/ |
the post we measured |
| Craft CMS | /blog/<slug> |
craftcms.com |
| Storyblok front end | /mp/<slug> |
storyblok.com |
The Wix /post/ segment and the Shopify /blogs/ segment are the two that bite. Neither carries information. Both are permanent for the life of the site, and both split your blog from the rest of your path structure. Our own product writes into customer CMSes and reads the URL the platform returns rather than the one we asked for, because the platform wins that argument every time. If you want control of that segment, the subfolder question is the one to settle before you publish a single post, not after.
A rename, caught in the act
Storyblok's page is the live demonstration. Ahrefs shows storyblok.com/mp/how-to-choose-cms-for-seo ranking ninth. Fetch it and you get:
$ curl -sI https://www.storyblok.com/mp/how-to-choose-cms-for-seo | head -3
HTTP/2 308
location: /mp/best-cms-for-seo
server: Netlify
The slug was renamed. The redirect is correct, it is a 308 rather than a 301 and Google names both, and sitemap-0.xml lists only the new URL out of 2,392 entries. Everything was done right. Google is still ranking the old URL, which is exactly what "expect temporary fluctuation" means in practice. That is the cost of a URL change on a page that already works, and it is the cost no listicle quotes.
Sitemaps: three defaults, and one that looks broken and is not
Measured the same day, at the origin of each platform. lastmod counts are occurrences in the file that actually holds the URLs, not in the parent index.
| Origin | robots.txt | Sitemap | URLs | lastmod |
|---|---|---|---|---|
| wordpress.org | 976 B | index, 3 children | 3 children | 3 |
| blog.hubspot.com | 5,831 B | 922,935 B single file | 7,958 | 7,958 |
| craftcms.com | 290 B | index, 39 children | 39 children | 39 |
| sanity.io | 317 B | index, 5 children | 1,031 in main | 1,031 |
| shopify.com | 3,803 B | sitemaps_list.xml |
548 | 548 |
| wix.com | 2,797 B | 2,359,811 B single file | 10,946 | 10,946 |
| webflow.com | 279 B | 302 off-domain | 9,074 in one child | 0 |
| squarespace.com | 2,379 B | 1,858,261 B single file | 6,369 | 6,282 |
| storyblok.com | 78 B | /sitemap.xml is 404 |
2,392 | 2,392 |
| rankability.com | 1,836 B | sitemap-index.xml |
285 | 0 |
Three of these would be flagged by any automated audit, and two of the three flags are wrong.
Storyblok's /sitemap.xml returns 404. Its 78-byte robots.txt declares /sitemap-index.xml instead, and that file resolves to 2,392 URLs with a lastmod on every one. The conventional path is a convention, not a requirement. Declaring the real path in robots.txt is the documented way to do it.
Webflow's /sitemap.xml is a 302 to another domain. It redirects to d3e54v103j8qbb.cloudfront.net/growth-seo/live/sitemap.xml, whose six children also live on CloudFront while listing webflow.com URLs. That reads like a misconfiguration and it is a supported pattern: Google's sitemap documentation describes cross-submission and says to make sure "the robots.txt file references the sitemap for that individual site", which webflow.com does. The 1,903 blog URLs are in there, inside a file named manual_sitemap_entries_sitemap0.xml.
The real finding is the third one. Webflow's 9,074-URL sitemap carries zero lastmod, and so does Rankability's. Google's position: "Google uses the <lastmod> value if it's consistently and verifiably" accurate. Omitting it is defensible, since a wrong date is worse than none. Shipping it on all 10,946 URLs, as Wix does, is the better default if the dates are real. Either way, "submitting a sitemap is merely a hint: it doesn't guarantee that Google will download the sitemap", and the Starter Guide adds that a sitemap "isn't required". A sitemap is not the control that separates these platforms. Indexing problems rarely start there.
One verbatim line from Shopify is worth more than any feature comparison. Inside allbirds.com's sitemap index, in an HTML comment: "The sitemap can not be edited manually, but is kept up to date in real time." That is a platform telling you plainly which control it has taken.
robots.txt, llms.txt, and the AI story the lists are selling
Three of the ranking pages promise a CMS ranking for AI visibility. Here is what the platforms themselves do about it, and what Google says it needs.
Every CMS in our set serves an llms.txt except Craft, which returns 404. The sizes are not close:
| Origin | llms.txt |
|---|---|
| blog.hubspot.com | 945,300 B |
| wix.com | 39,334 B |
| sanity.io | 30,133 B |
| webflow.com | 13,258 B |
| shopify.com | 12,638 B |
| agilitycms.com | 8,994 B |
| wordpress.org | 5,763 B |
| storyblok.com | 4,669 B |
| squarespace.com | 2,409 B |
| craftcms.com | 404 |
Three orders of magnitude of disagreement about a file that Google states it does not want. From its own documentation on AI features: "You don't need to create new machine readable files, AI text files, or markup to appear in these features", "There's also no special schema.org structured data that you need to add", and "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." We have written about what does seem to drive assistant citations, and it is not a file.
Two platforms do something more interesting. Both webflow.com and sanity.io serve a Cloudflare content signal in robots.txt:
User-agent: *
Content-Signal: ai-train=yes, search=yes, ai-input=yes
Cloudflare's policy defines search as "building a search index and providing search results", ai-input as "retrieval augmented generation, grounding, or other real-time taking of content", and ai-train as "training or fine-tuning AI models". Cloudflare's managed default for its own customers sets ai-train=no. These two platforms publish the opposite and opt in on all three, which is a content-licensing decision expressed in a text file, not an SEO setting. Cloudflare is candid about enforcement: "content signals express preferences; they are not technical countermeasures against scraping", and "Some companies might simply ignore them."
The check that fooled our own script
Our first pass flagged shopify.com as blocking all crawlers. The grep was for Disallow: / and it matched. Here is the context:
# robots.txt file for www.shopify.com
User-agent: GoogleDocs
Disallow: /
User-agent: AdIdxBot
Allow: */ppc/*
The block applies to one user agent named GoogleDocs. Rules in robots.txt belong to the group above them, and Google's documentation is explicit that only one group applies to any given crawler. A line-level check on this file produces a false alarm, and a false alarm in an audit costs you a morning. Read the group, not the line. This is the same trap as reading platform documentation instead of fetching the file: SEO automation is worth having, and only if the checks it runs are correct.
One last thing we found while measuring Shopify. Allbirds' sitemap index contains sitemap_agentic_discovery.xml, and that file holds exactly one URL:
<url>
<loc>https://www.allbirds.com/agents.md</loc>
<changefreq>weekly</changefreq>
</url>
A sitemap whose only job is to point crawlers at an instructions file for AI agents. It is new, it is not SEO, and it is the kind of capability that should influence a platform choice more than a plugin count.
Audit your own CMS in ten minutes
Replace the URL with one of your own published posts. Nothing here needs an account or a tool.
URL="https://yourdomain.com/blog/some-post"
# 1. Title, description and canonical, as the server sends them
curl -sL "$URL" \
| grep -oE '<(title|meta|link)[^>]*>' \
| grep -iE 'name="description"|rel="canonical"|<title'
# 2. Structured data: how many blocks, and which types
curl -sL "$URL" | grep -o 'application/ld+json' | wc -l
curl -sL "$URL" | grep -oE '"@type"[[:space:]]*:[[:space:]]*"[^"]+"' | sort -u
# 3. Body words the crawler gets without running JavaScript
curl -sL "$URL" | sed -e 's/<[^>]*>/ /g' | tr -s '[:space:]' '\n' | grep -c .
# 4. Sitemap and lastmod coverage
curl -s https://yourdomain.com/robots.txt | grep -i sitemap
curl -s https://yourdomain.com/sitemap.xml | grep -o '<lastmod' | wc -l
# 5. Does an old URL still redirect?
curl -sI https://yourdomain.com/blog/an-old-slug | head -3
Read the results against four thresholds, not against a vendor list:
- A meta description between roughly 110 and 155 characters, written per page. Four of our twelve are outside that range, including one with none.
- At least one JSON-LD block whose
@typematches the page. Five of twelve fail this on a blog post. - A raw word count within shouting distance of what you wrote. If
curlreturns 200 words on a 2,000-word article, your body is client-rendered and that is the finding. - A canonical that equals the URL you fetched, after redirects.
If all four pass, your CMS is not your problem and migrating will not help. If step 3 returns almost nothing, you have the one failure that genuinely justifies changing platforms.
When moving CMS is worth it, and when it is not
Changing CMS is a URL event, a redirect event and a re-crawl event at the same time. Google's Starter Guide sets the expectation on timing: "Some changes might take effect in a few hours, others could take several months." No one can guarantee what your traffic does on the other side, and anyone who quotes you a figure is selling.
Reasons that justify it, from the controls above:
- Your body text is not in the HTML. Nothing you write ranks if a crawler cannot read it. This is the only emergency on the list.
- You cannot set a canonical or a per-page title. Rare now, real on older hosted builders and on some custom builds.
- You cannot create a redirect. Every future URL decision is now permanent.
- You cannot export your content. The lock-in is the cost, and it compounds.
Reasons that do not justify it:
- A list said another platform is better for SEO. Six of the ten results here are written by one of the entries.
- Your HTML is large. Wix ships 2.6 MB and ranks on page one for queries we have measured.
- You want AI visibility. The AI-cited sample above is 44% sites with no detectable CMS and 24% WordPress. Platform is not the variable.
- Your current platform has no
llms.txt. Google says it does not need one.
The honest version of the recommendation is dull. If you are on WordPress, Craft, Ghost or Drupal and you can edit templates, you already have every control. If you are on Shopify or Wix, you have inherited a URL prefix and you should decide whether you care enough to move before you have 200 posts, not after. If you are picking a headless CMS, note that you are picking a storage layer and the real decision is the front end: Next.js gives you all eight controls and also makes all eight your job.
Then the work starts, and it is the same work on every platform: pick queries you can actually win, write pages that answer them better than what ranks, and keep publishing. That is what a content strategy is, and no platform does it for you. If you want the short version of where to spend the first month, our small business SEO checklist and the question of how long this takes cover it.
What the ranking pages get wrong
The deep-dive, for completeness, since the four readable results are what a searcher currently gets.
Agility CMS, position 3, DR 74, 3,780 words. Clean structure, SEO advantages and limitations per platform, a comparison table. Agility is the first entry and there is a section headed "Why Agility Is Perfect for Enterprises". Nine of its ten entries are headless or enterprise systems: Agility, HubSpot, Contentful, Strapi, Storyblok, Prismic, WordPress, Drupal, Sitecore. Nowhere does it measure anything, and nowhere is there a reader who owns one site.
Rankability, position 4, DR 75, 2,361 words, 61 backlinks. The best page in the SERP and the only one with original data, a documented detector, a stated accuracy figure and a limitations section. Two problems. Its trend table presents WordPress going from 49.9% to 20.0% as a change of "-29.9" while the methodology note says the baselines used a different detector and should be read "directionally", so the headline number is partly an artifact of its own instrumentation. And the audience is stated outright: a section titled "Best CMS for agencies", advice about "most agency clients", a conclusion that "Agencies win by remaining platform-agnostic". If you own one site, this page is about someone else's job.
Storyblok, position 9, DR 88. Ten platforms, Storyblok first, a section titled "Still wondering which CMS is best for SEO?" The checklist is sound and the evaluation questions at the end ("What questions should you ask a CMS vendor about SEO?") are genuinely the right questions. It is the page whose own slug was renamed mid-flight, which is a better lesson than anything in its body.
First Line Software, position 8, DR 52, zero backlinks, 1,763 words. Ranks eighth on a 1,416-search query with a page shorter than this section of our article, a Generator tag reading Site Kit by Google, no Article markup, and six FAQ questions doing most of the work. We say this with no superiority: it ranks, we do not, and the bar in this SERP is low enough that thin pages clear it.
What none of the four does is put a command in front of the reader. That is the gap, and it is why the top result is a Reddit thread. When every published answer is a vendor list, people go ask strangers instead. For the record, the Reddit thread at position 2 (DR 95, 1,026 estimated visits, 4 backlinks) is unreadable from our runner, so we are not characterising its contents.
FAQ
What CMS supports SEO best?
All of them, and that is the real answer. Measured on twelve vendor blog posts, eleven set a canonical correctly and eleven ship a meta description. The differences are in defaults rather than ceilings: WordPress.org's own post carries no meta description and no structured data, while Wix's carries six JSON-LD blocks. Pick on the eight controls in this article, then check your own site rather than a list.
Which CMS is best for SEO if I am starting from scratch?
Pick on URL control and on whether you will edit templates. If you want full control and have someone who can code, a framework such as Next.js or a traditional CMS like WordPress, Craft or Ghost gives you everything. If you want to never touch code, Squarespace emitted a complete set in our measurement, and Shopify or Wix will cost you a fixed /blogs/ or /post/ path segment forever. Decide on that segment before you publish.
What is the best headless CMS for SEO?
The question does not resolve, because a headless CMS does not render your pages. Contentful, Sanity and Storyblok hand JSON to a front end, and the front end owns the canonical, the markup, the sitemap and the render mode. Storyblok's own site is served by Netlify. Judge the front end, not the content API.
How do I migrate a CMS without losing SEO?
Keep every URL identical if you possibly can, and where you cannot, serve a permanent redirect from old to new. Google recommends "HTTP permanent redirects if possible, such as 301 and 308" and says to keep them "generally at least 1 year", while warning to "Expect temporary fluctuation in site ranking during the move". Results vary, and the Storyblok page in this article is still ranking on its pre-rename URL weeks later, so plan for a gap rather than a switch.
Does a CMS affect whether AI assistants cite my site?
Not measurably. In Rankability's sample of 209 AI-cited sites, 44.0% ran no detectable CMS, 24.4% ran WordPress and 15.8% ran Next.js, which mostly reflects what kind of site gets cited rather than what software it uses. Google states there are "no additional requirements to appear in AI Overviews or AI Mode". What varies is whether your content answers the question, which is the subject of answer engine optimization and LLM SEO.
Which ecommerce CMS is the best for SEO?
Shopify appears on 6.7% of top-ranking domains in the study above, more than Drupal, Webflow and Wix combined, and it takes the /blogs/ prefix as payment. WooCommerce on WordPress gives you the URL back and hands you the maintenance: see WooCommerce SEO and our ecommerce SEO checklist. Note that Shopify's own blog post marked itself up as a PodcastEpisode with no article schema, so check your store's templates rather than assuming the platform did it.
Get cited by ChatGPT. Rank on Google.
You found this article through search. That is the whole product.
- One researched article a day
- Published on your own domain
- Keywords checked against live results
Get cited by ChatGPT. Rank on Google.
You found this article through search. That is the whole product.
- One researched article a day
- Published on your own domain
- Keywords checked against live results
