Headless CMS or website builder? When Sanity + Astro beats Framer and Webflow

Most B2B marketing sites are better served by a website builder: Framer if design speed and visual editing matter most, Webflow if your team already knows it or needs a larger CMS. Going headless, with Sanity for content and Astro or Next.js for the pages, pays off under specific conditions: several languages (especially right-to-left), content reused across pages or products, hundreds of structured pages, strict control over HTML and structured data, real integrations, or a site that is part of a product. It also needs a developer who owns the code, because every design change becomes a code change.

Our bias, stated once: we build on all of these, we're an Official Framer Expert, and our own site, agr.studio, moved from Framer to Astro + Sanity in October 2026. Prices are list prices as of October 2026, linked to each vendor.

What does "going headless" actually mean?

A headless website is three systems instead of one. Content lives in one: a headless CMS such as Sanity, where your team writes pages, posts and case studies in forms. The pages are built by another: a framework such as Astro or Next.js, where a developer writes the templates that turn that content into HTML. The finished site is hosted on a third: Vercel, Netlify or Cloudflare. A builder like Framer or Webflow is all three in one product: you design, write and publish from one screen, and hosting comes with the plan.

That split is the trade. Each part is yours to choose and control, and someone has to hold the three together.

How do Framer, Webflow and Sanity + Astro compare?

When is a website builder the better choice?

When the people who change the site are marketers and designers, not developers. In 101 client conversations we coded between January and October 2026, platform choice came up in 46% and Framer was named in 34. The reasons teams gave for picking it: the marketing team can edit without a developer, hosting is included, and the editor feels like Figma.

A builder fits when:

• Layouts change often, and a designer wants to ship sections the same day.
• There is no developer in-house or on retainer.
• You run one or two languages. We built Baladi, a bilingual Hebrew and English site, on Framer in three weeks.
• Launch speed matters. Our usual quote for a standard B2B site is 3 to 5 weeks.

For Framer against Webflow, see Framer vs Webflow for B2B; for every option side by side, the 2026 platform comparison.

When does going headless pay off?

When at least two of these are true and someone can own the code. One alone rarely justifies three systems.

1. Several languages, especially with right-to-left among them. Two languages, even Hebrew and English, are fine on a builder. Past that, locale fees add up, while Sanity charges none and handles translation per document or per field with its own plugins. The fee matters less than the work: you want per-locale metadata generated rather than typed by hand, and with Hebrew or Arabic, direct control of the mirrored layout.
2. Content reused across pages or products. A customer quote on the homepage, three industry pages and a product page should be one record, edited once. A headless CMS treats content as structured data any page, product or channel can pull. Weighing Sanity against WordPress instead? See Sanity vs WordPress.
3. Hundreds of pages with structure. Integration directories, glossaries, docs, locations. Framer Basic is fixed at 30 pages; Pro allows 150, then $20 per 100 up to 700. On headless, pages are generated from content types.
4. Strict control over HTML, speed or structured data. Astro prerenders the whole site by default, and its components render to static HTML with no client-side runtime. Every tag and every JSON-LD block is yours. On Framer, structured data goes in through custom code.
5. Integrations. Forms that write to your CRM, product data from an API, endpoints that AI assistants can call.
6. The site is part of a product. Logged-in areas, app-like features, heavy personalisation. That points to Next.js rather than Astro; see Astro vs Next.js.
7. A developer in-house or on retainer. Without one, every design change waits in a queue.

What does headless cost that the pricing pages don't show?

People and process. Platform fees are close to a builder's; the real cost is that design lives in code.

• More moving parts. Three vendors, three bills, three logins. Vercel's free Hobby plan is "restricted to non-commercial personal use only", so a company site needs Pro at $20 a month with one deploying seat.
• A developer for every design change. A new section type, a site-wide spacing change, a new button style: each is a code change that gets reviewed, built and deployed. Editors only get the blocks a developer has built.
• Build and deploy. Our own build takes one to two minutes for about 300 pages. Quick, but a canvas has no such step.
• Preview setup. Sanity includes Visual Editing and its Presentation tool on all plans, with an official Astro quickstart, but a developer has to wire it in. Until then, editors fill in forms without seeing the page.
• Upgrades. Astro 7.0 shipped on 22 June 2026 and Sanity Studio v6 on 11 June 2026. Each major is developer time a builder never bills you for.

What do builders cost that the plan price hides?

Seats, locales and limits. The plan pays for the site; people and languages are billed on top.

• Editor seats. Framer caps edit access at 10 seats on Basic and Pro. Since 5 October 2026, Webflow's legacy Editor "can no longer be accessed", and anyone editing content needs a seat on the site's Workspace. Editors moved over from the old Editor got free seats, and sites kept in an agency's Workspace get free client seats; a new editor on your own Workspace is a paid seat. Teams adding editors have a new budget line.
• Locale add-ons. Four added languages on Framer is $80 a month. Webflow's cheaper Localize tier stops at 3 locales; the $29 tier goes to 10.
• Plan limits. Framer Basic stops at 1,000 CMS items; Webflow Basic has no CMS, so a blog means Premium.
• Export limits. Webflow exports HTML, CSS and JavaScript only on paid Workspace plans, without CMS content, form processing, localized content or code components; collections come out as CSV. Framer's help centre contradicts itself on exporting a published site; CMS content can be exported through plugins as CSV or JSON. More in who owns your website after launch.
• Code-component SEO caveats. If a Framer code component skips pre-rendering, "the component's content won't be accessible to search engines", and Framer says there is "currently no way to make overlay content crawlable". Webflow server-renders code components by default, but that can be switched off. Google's advice applies everywhere: pre-render, because "not all bots can run JavaScript".

Who edits what on each platform?

On a builder, whoever holds the right seat can change anything they can see. On headless, editors change content and developers change the system it flows into.

What does a year cost for a realistic B2B site?

Platform fees land within a few hundred dollars of each other. Developer time decides the total.

Assumptions (ours, so the totals are estimates): about 60 pages plus 300 posts and case studies; English plus one language; an owner, a designer who changes layouts and three content editors; traffic inside included bandwidth; yearly billing where offered. Webflow on the free Starter Workspace, owner's full seat included. Sanity with five seats (owner, three editors, developer), the designer as a free Viewer. Excludes the build, domain, tax and third-party apps.

On a builder, developer hours cover custom code, embeds and integrations. On headless, they assume one new or changed section a month at 4 to 8 hours each, plus upgrades and build fixes. Use your own rate.

Languages move the numbers most: four more locales add $960 a year on Framer and nothing on Sanity. A Sanity Growth seat costs more than a Framer content editor, though. Sanity Free would cut its line to $0, but Sanity describes it as "for individuals and hobbyists", with only Administrator and Viewer roles. If you already employ a developer, the headless hours are time you have, not a new bill.

Why did we move our own site from Framer to Astro + Sanity?

Because agr.studio meets several of the headless conditions, and we are the developer. Full story: Framer to Astro + Sanity. In short:

• History. Framer until 16 October 2026. A Next.js + Sanity rebuild started in January and never launched. In October we rebuilt on Astro + Sanity, hosted on Vercel. Astro, because it's a content site: every page is prerendered; only the embedded Studio and a few endpoints run on demand.
• Scale. About 300 indexable pages: English plus a full Hebrew right-to-left mirror, with hreflang between them. About 80 redirect rules ported from Framer paths (91 today, after later renames).
• What control bought us. JSON-LD on every page type, /llms.txt, a Markdown version of every page, a form endpoint with an OpenAPI file so AI assistants can send an inquiry, and the live AI check. The new build's layout shift on load went from a CLS of 0.19 to 0 before launch (font metric overrides, reserved image space).
• What we gave up. On-canvas editing and instant publishing by a designer. Design changes now go through a build and a deploy. Content edits stay self-serve in Sanity.

AI coding agents (Claude Code) did much of the implementation under our direction and review. We still build client sites on Framer.

What we've seen on client projects

• Editing without a developer is why teams pick a builder. CMS and self-editing came up in 20% of our conversations. "Can our team add case studies ourselves?" is a yes on all three options. Adding a new kind of section is where headless, by design, says no.
• "We have no Figma; the website is our design system." For one team, a new page took about two weeks. Our plan was a CMS plus documented components, to roughly halve that. The fix was components, not a platform switch. See a CMS your team can run.
• Multilingual sites grow the hidden work. A cybersecurity SaaS migrating five languages asked for "SEO and LLM optimisation": structured data was missing, and per-locale metadata had to be generated, not hand-typed. We quote about 15 business days for a five-language migration with QA, with every old URL mapped in every language.

Which side are you on? A yes-or-no checklist

1. Will you publish in three or more languages, especially with a right-to-left one among them?
2. Does the same content (quotes, product facts, case studies) need to appear on many pages or in other channels?
3. Will you have more than a few hundred structured pages, such as integrations, docs or locations?
4. Do you need control over HTML, structured data or performance that your builder doesn't expose?
5. Does the site need to talk to other systems: a CRM, product data, an API, AI assistants?
6. Is the site part of your product, with logins or app-like features?
7. Do you have a developer in-house or on retainer who will own the code?
8. Can your designers accept that design changes go through a developer?

Yes to 7 and 8, and to at least two of 1 to 6: headless is worth pricing out (how we build headless sites). A no on 7 or 8: stay on a builder, whatever else is true. Either way, pick the platform after the requirements, not before; that's platform-agnostic web design.

Not sure where you stand? Run the free AI check on your current site, or tell us about it.