Webflow to Astro Migration: The Complete 2026 Guide

Webflow to Astro, in Short

For a 50 to 100 page marketing site, budget $5,000 to $25,000 and 4 to 8 weeks. The work is a rebuild, not a conversion. You take an inventory of what exists, export what Webflow will give you, pull the CMS through the Webflow Data API, rebuild the design as Astro components, map every old URL to a new one, and cut over DNS.

The mechanism people are buying is simple: Astro ships zero JavaScript to the browser by default, per Astro's own documentation, and adds client-side JavaScript only where you explicitly opt in with a client directive. Webflow serves its own runtime, webfont loader and interaction scripts on every page whether a given page uses them or not. That is an architectural difference, not a benchmark, and it is the honest version of the performance argument.

Three things this page deliberately does not do, because they have their own homes:

• Feature-by-feature platform comparison lives in Astro vs Webflow.

• CMS field mapping, schemas, reference fields and rich text conversion live in migrating Webflow CMS content to Astro.

• The general case for Astro as a framework, whether or not you are on Webflow, lives in why Astro suits marketing sites.

This page assumes you are deciding whether and how to move.

Should You Move At All?

Most Webflow sites should stay on Webflow. The signals below are the ones that actually predict a migration paying off. One of them on its own is rarely enough. Three or more together usually is.

1. You are embedding more code than you are designing. Count the custom code embeds in your project. Analytics, chatbots, CMS filters, site search, cookie consent, A/B testing, form validation. Past roughly five, the visual builder has stopped being the thing doing the work, and each embed is an untested, unversioned dependency that nothing in your stack can review.

2. Seats cost more than the site. A full Webflow seat is $39 per month on yearly billing and a limited seat is $15 per month on yearly billing, on top of the site plan. Every person who needs to touch the site adds a recurring line item. That is the durable Webflow cost story, and unlike plan pricing it did not get cheaper in 2026.

3. Your included bandwidth went down. Webflow's May 13 2026 change merged the CMS and Business plans into Premium. Sites that were on Business and moved to Premium saw included bandwidth drop from 100GB to 50GB. This is not universal: sites on the old CMS plan saw their limits go up, so check which plan you were actually on before you treat it as a reason to leave.

4. You depend on something Webflow retired. Webflow's User Accounts product, formerly Memberships, was sunset on 2026-01-29 and is no longer available on the sites that used it. If you built gated content on it, you already have a rebuild on your hands and the only open question is where it lands. The legacy Editor and whitelabeling end on 2026-08-04, though that one is narrower than it sounds: the replacement editor role is free, the migration ran automatically from 2026-05-04, and nothing about the live site changes. Whitelabeling is the genuine loss, and it only ever applied to Growth and Agency Workspaces.

5. You need the things a git repository gives you. Branching, pull request review, preview deployments per branch, and the ability to revert a bad change in one command. If two or three people are changing the site every week, the absence of a review step is the recurring cost, not the subscription.

6. You want an AI agent to do real work on the site. Be precise here, because the common version of this claim is wrong. Webflow ships an official MCP server at mcp.webflow.com/mcp, listed in Anthropic's connector directory since early 2026, and it can read your site structure, bulk read and write CMS items, create Collections and fields, update page SEO metadata and publish a site. The real limitation is the shape of that access: a partial, rate-limited (60 requests per minute on Starter and Basic, 120 on CMS, Ecommerce and Business), unversioned interface with no build, no test and no diff to review. Asset uploads to the media library are not exposed, and Interactions still have to be built by hand in the Designer. In a repository, an agent gets the whole artifact, and can prove its change compiles before you look at it.

What is not a reason: the CMS item ceiling. Webflow's May 2026 change raised item limits rather than lowering them, to 20,000 items on Premium and Team. Any article telling you to migrate because you are about to hit a 2,000 item wall is describing a limit that no longer exists.

When Not to Migrate

This is the section that makes the rest of the page worth trusting. Stay on Webflow if:

• Your site is a brochure that works. Five to fifteen pages, no CMS to speak of, generating the leads you need. There is nothing here for a migration to fix, and you will spend $5,000 or more to end up with the same site plus a maintenance burden.

• Nobody on your side can own a codebase. A migration hands you a repository. If there is no developer, no contractor on retainer, and no appetite to hire one, you have swapped a subscription you understand for a dependency you do not.

• A designer without engineering support ships all your pages. Webflow's actual advantage is that a designer can put a new page live without a build step or a deploy. If that is your workflow and it works, a code migration removes the thing you are paying for.

• You are mid-rebrand or mid-redesign. Migrate the site you are going to keep. Doing both at once means every launch bug is ambiguous: nobody can tell whether the migration broke it or the redesign changed it.

• Ecommerce is the business. Webflow Ecommerce is not sunset and nothing is changing for those plans at this time. Moving a live store means rebuilding checkout, tax, inventory and fulfilment, which is a different and much larger project than moving a marketing site.

• Your traffic is small and your rankings are everything. Any migration puts rankings at risk during the cutover. If organic search is the only channel and the upside is a slightly faster site, the risk-adjusted return is poor.

What Webflow Actually Hands You On the Way Out

Plan around this before you scope anything, because the gap between what people assume export includes and what it actually includes is where budgets go wrong.

Code export is gated on the Workspace plan, not the site plan. Core, Growth, Freelancer and Agency Workspaces include it; the free Starter Workspace does not. Webflow's own wording is that site plans do not include the ability to export code.

What you get: HTML, three CSS files, the JavaScript folder, images from the Assets panel, and *empty* Collection template pages.

What is excluded: CMS content, meaning every Collection item and every Collection list. Also User Accounts, Ecommerce, password protection, form submission processing, reCAPTCHA, site search, non-primary locales and code components.

Two details worth correcting, because both are commonly stated backwards:

• Interactions do export. IX2 interactions come across and function through the exported JavaScript folder. Lottie animations need to be served from a web server rather than opened as local files. The claim that Webflow interactions cannot be exported at all is wrong.

• The CMS is the part that does not come with you. That is why the export is a starting point for markup and assets, and the Data API is how the content actually moves. The field-by-field mechanics of that, including how rich text and reference fields convert into Astro content collections, are covered in the CMS migration guide.

The Migration, Step by Step

1. Inventory before you open an editor

Document every page with its URL and its traffic, every Collection with its fields and its references, every form and where submissions land, every custom embed, every interaction, and your current indexed-URL list. This is the phase that decides whether the project takes four weeks or ten, because everything discovered later is rework.

Export your sitemap from Project Settings, SEO, Sitemap, and reconcile it against the Designer page list. Orphaned pages, old campaign landers and utility pages hide in the gap between the two.

2. Decide what does not come along

A migration is the cheapest moment you will ever have to delete pages. Audit against traffic data and retire the campaign pages that ran once in 2023 and the four losing variants of the pricing page. Fewer templates is the single biggest lever on cost.

3. Build the redirect map first, not last

Every URL that changes needs a 301. Build the map as a data file before the rebuild starts, so it can be tested as pages land rather than assembled under launch pressure. In Astro this lives in astro.config.mjs or at the host. Test every entry against the live old URL list before cutover, and keep the old sitemap to diff against the new one.

4. Rebuild the design as components

Do not try to convert the exported HTML into a maintainable codebase. Webflow's output is deeply nested divs with generated class names, and cleaning it is slower than rebuilding. Use it as a reference for spacing, breakpoints and copy.

Build a base layout in src/layouts/, then rebuild each repeated section as its own component: hero, feature grid, testimonial block, footer CTA. What was a symbol in Webflow becomes a single .astro file with scoped styles. AI coding tools are genuinely useful in this phase, because they can work from the exported markup and a screenshot at the same time.

5. Move the content

Pull Collections through the Webflow Data API, import in dependency order so referenced Collections exist before the Collections that point at them, and validate against a schema on the way in. Astro type-checks content collections at build time, so a field that fails to map surfaces as a build error rather than as a blank space on a live page. The detail is in the CMS migration guide.

6. Rebuild what does not transfer

• Forms. Webflow's form processing does not leave Webflow. Replace it with a form service or an Astro server endpoint, and confirm the receiving inbox, CRM or automation before launch, not after.

• Interactions. Exported IX2 works, but carrying Webflow's runtime into a new codebase to keep it working is usually the wrong trade. Rebuild scroll and hover effects with CSS animations and IntersectionObserver, and reach for an animation library only where the effect genuinely needs it.

• Gated content. If it was built on User Accounts, it needs a new provider regardless, since that product is gone.

• Third-party tooling. Search, CMS filtering and similar add-ons each need an equivalent, or a decision to drop them.

7. Launch and watch

Publish, submit the new sitemap to Search Console the same day, and monitor coverage and rankings for the following two to three months. Redirect errors show up in Search Console before they show up in revenue.

What It Costs

$5,000 to $25,000 for a 50 to 100 page site. That is the band, and where a given project lands inside it is decided by five things:

1. Unique templates, not page count. Two hundred blog posts on three templates is a smaller job than fifteen pages that are each bespoke.

2. CMS complexity. A flat blog is straightforward. Interlinked Collections with reference fields, conditional content and multiple locales are not.

3. Interactive work. Scroll-triggered animation, custom JavaScript and third-party widgets are each rebuilt by hand.

4. SEO exposure. A site with meaningful organic traffic needs a far more careful redirect map and a longer verification pass.

5. Timeline. Compressed schedules cost more, because parallelising phases costs coordination.

Doing it yourself trades money for calendar time and risk. The parts that go wrong are rarely the visual rebuild. They are the redirect map, the CMS reference graph, and the forms nobody remembered were wired into a CRM.

Hiring a freelancer works when the scope is genuinely small. The recurring failure is scope: the quote covers the visible rebuild, and then the CMS import, the redirect mapping and the edge cases arrive as change requests.

Costs that get missed in most estimates: content cleanup, because Webflow Collections accumulate inconsistent formatting and broken references; DNS and any email routing you run through Webflow; replacements for third-party add-ons; getting editors comfortable with a new CMS interface; and two to three months of post-launch SEO monitoring.

What you pay after you move

Be careful with the "hosting is free" story, because it usually is not.

• Vercel Hobby is non-commercial only. Vercel's terms define commercial usage to include a paid employee or consultant writing the code, and receiving payment to create, update or host a site. A client site on Hobby is a terms violation. Vercel Pro is $20 per user per month, and that is the realistic floor there.

• Cloudflare Pages has unlimited bandwidth and no commercial restriction, which makes it the one honest zero-cost path. Pages Functions count against the Workers free allowance at 100,000 requests per day.

• Netlify's free tier roughly halved in April 2026, to around 15GB per month with no overage purchase, so it is a poor fit for anything with real traffic.

Realistically: $0 to $20 per month on a static-first stack hosted on Cloudflare Pages, or roughly $20 to $50 per month on Vercel with a managed database. Price a hosted CMS separately. Sanity's free tier covers 20 seats and 10,000 documents, with Growth at $15 per seat per month. Storyblok's free tier is a single seat, which is usually a blocker for client handoff, and Growth is $99 per month. Payload is MIT-licensed and free as software, but it needs a database you pay for, realistically $8 to $30 per month, and Payload Cloud has paused new deployments, so hosting is your problem. Payload was also acquired by Figma in June 2025, which is worth knowing if you are assessing vendor risk.

Timeline

• Small marketing site, under 20 pages, no CMS: roughly 2 weeks.

• Typical 50 to 100 page site with a CMS: 4 to 8 weeks across audit, rebuild, content migration, QA and launch.

• Deeply referenced Collections, gated content or heavy interaction work: 8 to 12 weeks or more.

The phases that should never be compressed are redirect mapping and QA, because those are the two that protect your rankings on launch day.

Getting It Scoped

Every number above is a starting range, not a quote. What turns it into a real figure is a look at your actual Collection references, your Interactions, your integrations and your indexed URL list. Reach out through migratelab.com for a migration review and you get a scope, a timeline and a fixed price before you commit to anything.