Strapi was named in 5 of the 6 answers we collected, the most of any headless CMS, when we asked ChatGPT and Gemini which one to use. Hygraph, Sanity and Storyblok tied just behind, each named in 4 of the 6 answers. No other product cleared three.
Here is what we did, and it is repeatable. On 7 August 2026 we put six buyer questions about headless CMS software to ChatGPT and Gemini, gave each answer one live web search, and wrote down every product named and every source domain cited. Six questions, six answers, 23 distinct product labels and 30 distinct source domains. Every number on this page comes from that one run and nothing else.
The head of the list was stable and commercial. Strapi (5 of 6), Hygraph (4 of 6), Sanity (4 of 6), Storyblok (4 of 6), Directus (3 of 6) and Prismic (3 of 6) were the only six products named in at least three answers. The other 17 labels were each named in one or two answers, so the middle and tail of this market is where the recommendations get unpredictable.
How the shortlist changed across the six questions
The most useful thing in this data is how much the shortlist moved when we changed one word in the question. The bare "best headless cms software" question returned eight established, commercial names: Storyblok, Sanity, Strapi, Contentstack, Squidex, Hygraph, Directus and Prismic. Ask instead for the "best free" option and five of the six names that came back had not appeared in the "best" answer at all.
The "best free headless cms software" answer named six products: Decap CMS, Sveltia CMS, ChiliCMS, Payload, TinaCMS and Strapi. Only Strapi carried over from the bare "best" list. The other five are git-based or open-source tools that the commercial-sounding question never surfaced.
"Most affordable headless cms software" produced a third, different mix of eight: Storyblok, Strapi, Payload, TinaCMS, Contentlayer, ChiliCMS, Sanity and Decoupla. It shared four names with the free answer (Strapi, Payload, TinaCMS and ChiliCMS) but only three with the bare "best" answer (Strapi, Storyblok and Sanity). Free and affordable are treated as closer to each other than either is to best.
Eleven products were each named in exactly one of the six questions, which is where the fan-out shows most. "Best headless cms software for small teams" alone produced five of them: Adapto CMS, b10cks, BaseHub, Decoupled.io and TinyCM, none of which appeared in any other answer. Squidex showed up only under "best", Contentful only under "what should I use", Decap CMS only under "best free", Contentlayer and Decoupla only under "most affordable", and Umbraco only under "recommendations".
One brand shows the fan-out inside its own name. The assistants wrote "Payload CMS" when answering "what should I use" and "recommendations", but "Payload" when answering "best free" and "most affordable". That is the same product in 4 of the 6 answers, yet because it was labelled two different ways it never shows higher than a pair of 2s. We keep both rows in the table below rather than merge them into a number the run did not actually produce.
The source map
Answers in this category are fed mostly by vendors' own websites, not by community forums. Of the 30 source domains, 17 are a CMS product's own home page, and 16 of those 17 belonged to a product that got named in the answer their page fed. The one exception was firecms.co, whose product FireCMS was cited as a source but did not make the recommendation.
The single most-cited source was strapi.io, which fed 4 of the 6 answers, more than any other domain, and Strapi was also the most-named product at 5 of 6. That pairing is the key insight on this page: the product whose own domain the assistants leaned on most was the product they recommended most. Storyblok's own site fed 2 answers and Storyblok was named in 4, and the same self-reinforcing pattern shows up smaller for Hygraph, Sanity, ChiliCMS and b10cks.
Community threads did no work here at all. Not one Reddit or Quora page appeared among the 30 sources. The closest thing to a community source was github.com, cited in 3 of 6 answers, and that is code repositories rather than discussion. Roundup and comparison blogs carried the rest: pantheon.io was cited in 3 answers, and softwareadvice.com, techsy.io, luckymedia.dev, tinyctl.dev and cxeverywhere.com in 2 each.
Classic review directories barely registered. G2 appeared once, via learn.g2.com in the "recommendations" answer. Software Advice appeared in 2 answers. Capterra did not appear at all. In a category people assume is ruled by G2 and Capterra grids, these six answers were built almost entirely from vendor pages and independent roundups instead.
What a headless CMS vendor would do to get named
If you sell in this category, your own website is the highest-leverage place to be, because 16 of the 17 vendor home pages in this source set belonged to a product that got named. Strapi did not top the list by being the biggest brand, it topped it because strapi.io was the most-retrieved source in the run, feeding 4 of 6 answers. Make your product pages the kind a live search will pull and quote.
Market share does not decide this, the source list does. Contentful, one of the best-known commercial names in headless CMS, was named in only 1 of 6 answers and its own domain never appeared as a source at all. Meanwhile Adapto CMS, b10cks and TinyCM each got named once, purely because their own home page (adaptocms.com, b10cks.com, tinycm.com) happened to be retrieved. Being obscure did not keep them out, being un-retrieved kept Contentful down.
After your own domain, target the exact roundups these answers used. Pages on pantheon.io (3 answers), softwareadvice.com (2), techsy.io (2), luckymedia.dev (2), tinyctl.dev (2) and cxeverywhere.com (2) are the ones that carried the mid-pack names into the recommendations. Getting listed and described accurately on those specific pages is how a product travels from unknown to named.
The full list, counted
| Product | Named in | Share |
|---|---|---|
| Strapi | 5 of 6 | 83% |
| Hygraph | 4 of 6 | 67% |
| Sanity | 4 of 6 | 67% |
| Storyblok | 4 of 6 | 67% |
| Directus | 3 of 6 | 50% |
| Prismic | 3 of 6 | 50% |
| ChiliCMS | 2 of 6 | 33% |
| Contentstack | 2 of 6 | 33% |
| Payload | 2 of 6 | 33% |
| Payload CMS | 2 of 6 | 33% |
| Sveltia CMS | 2 of 6 | 33% |
| TinaCMS | 2 of 6 | 33% |
Every product named in at least 2 of the 6 answers. Payload and Payload CMS are the same brand under two labels, kept separate because that is how the run recorded them. Another 11 products were each named once.
Where Glotier fits
Glotier does not sell a headless CMS, so we are correctly absent from all six of these answers, and we are not going to pretend otherwise. The value here is the method, not our placement.
This is exactly the check we run for a customer's own category: the same six-question fan-out, the same record of which products get named and which pages fed each answer, so you can see whether your product shows up when a buyer asks an assistant what to use, and which sources you would need to be on to change that. It is free, needs no account, and takes about a minute.