Framer fits a design-led marketing team, whilst WordPress suits a publishing and plugin workflow. Astro plus a headless CMS is for a team ready to own the build and hosting. Keep Webflow on the shortlist if its editor and content model fit the job and one integration can solve the problem. Then weigh each route against the content, forms, memberships and URLs that must survive a move.
Match the route to the requirement that is failing
Start with what is actually broken, not what merely feels dated. If editors are happy and the only failure is login, a full rebuild is a heavy answer to a small question.
The table below matches each route to the situation it suits, and to a trial task. That trial task is deliberate: an editor alters a representative content page and publishes it. Recreating a homepage screenshot proves nothing about the workflow your team will depend on for years.
| Route | Choose it when | What carries over | What needs rebuilding | Trial task |
|---|---|---|---|---|
| Framer | A design-led marketing team wants to edit and publish on a visual canvas | Visual editing habits, plus collection content through CSV import (a feature Framer lists) | Pages, collections and references, forms, gated content, localization setup | An editor opens a representative article, changes a heading and a reference field, previews it, then publishes |
| Self-hosted WordPress | Publishing volume, user roles, scheduling and a plugin ecosystem matter most | Content once mapped into posts, pages and custom fields; media once downloaded | Theme and templates, forms, membership, SEO plugin setup, redirects | An editor drafts a post in the real template, previews it, schedules it, and a lower-role user tries to publish it |
| Astro plus a headless CMS | A team with engineers will own the build, hosting and preview pipeline | Exported markup as a reference, CSV content, downloaded assets | The whole front end, editor interface, previews, forms, search | An editor changes an entry in the CMS, sees a preview, publishes, and the live page updates |
| Stay on Webflow with a focused integration change | Editors like the workflow, the content model works, and the gap is membership or one other integration | Pages, CMS, hosting, URLs | The login, the gate and its link to billing | A test member signs up, reaches a gated page and logs out, while an editor publishes a different page |
Astro is not a turnkey visual builder, so do not judge it like one. It is a framework for people who write code, which is precisely why the Astro CMS guide exists: content lives somewhere else and you connect it. If you want the longer argument, our headless CMS page covers it, and Astro vs Next.js covers the framework choice.
Before you pick anything, write down what the export won't hold:
- Every CMS collection and the references between them
- Every form and where its submissions go
- Every protected or members-only page
- Every locale beyond the primary one
- Every external service wired to the site
- Every URL that earns traffic or links
What Webflow code export does and does not move
The word export promises more than it delivers. According to the official code export guide, paid Workspace plans can export a site's HTML, CSS, JavaScript and assets. A Site plan alone cannot. You open the Main menu, choose Code export, or press Shift + E, click Prepare ZIP and download the file.
Inside the ZIP you get a CSS folder, a JS folder, an images folder with everything from the Assets panel, and HTML files for all pages, including empty Collection template pages. The guide also warns that deleting the JS files can leave a page looking intact while failing to function. You cannot choose which parts get exported, and manual edits to exported files may break some Webflow components.
That gives you a usable starting asset, and a decent one. You can host it anywhere, with no attribution to Webflow and no Webflow plan needed for it to run on your own server. But it is static markup.
The guide lists what is not included and will not function:
- CMS content and functionality, including Collection items and lists
- User Accounts content and functionality
- Ecommerce content and functionality, including products and checkout
- Site and page password protection
- Form submission processing, plus reCAPTCHA elements
- Site search
- Localized content (only the primary locale's pages, elements and content are included)
- Code components
On an exported site, Collection lists render their empty state and Collection pages show no content bound to Collection fields. Webflow says its password protection only works on Webflow hosting. Rebuild and test access controls before placing previously protected pages on another host. The ZIP does not preserve Webflow's hosted login and password-protection features.
Forms and search do not work on exported sites either, which Webflow attributes to GDPR compliance. You need a third-party tool or an embedded form integration to collect submissions.
A CSV is data, not a CMS
Some underlying content can be backed up separately. The same guide describes CSV exports for CMS Collections, including localised Collection content, and Ecommerce Collections. But a CSV carries records, with no editing interface, no preview system, and no checkout. Its historical User Accounts export instructions do not establish that those records remain available after the completed sunset described below.
One further catch, noted by a third-party write-up from Figment: image and file fields in CSV exports come out as URLs pointing back to Webflow's asset hosting. Download those files before you cancel hosting, or your content points at dead links. That is a secondary source, so check it against your own export.
Three replacement routes, same five questions
Put the same questions to each route: who edits, how content relationships and previews work, who runs hosting and updates, which integrations you need, and how URLs and metadata survive. Vendor feature lists are descriptions rather than proof. The trial task from the table is where the real answer lives.
Framer
Who edits: Framer describes a visual editor where you edit and publish content directly on the canvas, with drafts and a preview before publication. That is a vendor claim, but it matches the design-led workflow this route suits.
Relationships and previews: Framer lists references, CSV import and built-in localisation on its CMS page. Test a representative collection and its relationships. Then check current plan limits against your actual content inventory before committing.
Hosting and integrations: The platform hosts for you, which is the appeal. Forms, gated content and anything unusual still require an integration you pick and test. I won't make a blanket claim about exporting a Framer site back out, because Framer's current documentation on that point needs project-level verification.
URLs and metadata: Mapped by hand. Same as every route.
Self-hosted WordPress
Keep two things apart. WordPress.org is software: GPL-licensed, no licensing fees, and you run it on hosting you choose. WordPress.com is a hosted service sold by a company, where what you can install and who patches it depends on the plan. Do not treat them as one decision.
Who edits: WordPress.org lists drafts, scheduled publication, revisions, public or private content and password-protected posts and pages as standard. It also describes user roles: administrators, editors, authors, contributors and subscribers. For an editorial team with several levels of access, that is a genuine fit.
Relationships and previews: Custom post types, taxonomies and metadata exist through code, per the same features page. Modelling Webflow reference fields takes planning, and probably a developer. Previews depend entirely on your theme.
Hosting and updates: Yours. WordPress's plugin documentation says plugins are PHP scripts, often developed by volunteers, and of varying quality. A plugin not updated since the latest core update may be incompatible or have unknown compatibility. Someone has to own that maintenance indefinitely.
Integrations, URLs, metadata: Plugin categories cover SEO, e-commerce, security, caching and data import and export. Redirects and metadata still need mapping. For the content modelling and editing work involved in a rebuild, see the headless CMS development page.
Astro plus a headless CMS
Who edits: Editors work in the CMS chosen for the project, which can include visual editing. Sanity's documentation describes live previews and click-to-edit when connected to the frontend. Don't assume Astro comes with either a visual canvas or a form-only interface; put that integration and an editor trial into the scope.
Relationships and previews: Structure comes from the CMS you choose. Previews are something your engineers build. Do not assume they arrive free.
Hosting and updates: Your team picks the host, the build process and the dependency updates. That is the engineering ownership I mentioned at the top. If nobody wants that job in two years, pick a different route.
Integrations, URLs, metadata: Everything is explicit. Forms, search and redirects are all things you wire up. The upside is control. The cost is that nothing happens unless somebody writes it.
Memberships after the User Accounts sunset
This is no longer a deadline; it is a completed change. As of November 4, 2026, Webflow's User Accounts sunset notice describes a retirement date of January 29, 2026, which has passed. The dedicated User Accounts APIs and webhooks went away on the same date, hitting any integration, workflow or custom development that depended on them.
Webflow's notice said sites still using User Accounts on that date would lose access to the feature and its data, with nothing migrated to the CMS. User Accounts pages became regular static pages, content gates were removed, and members could no longer log in on published sites. Do not plan around retrieving historical User Accounts data from Webflow now. Access was lost, so any recovery depends entirely on a CSV exported before the date.
The notice and the original December 11, 2024 announcement name Outseta and Memberstack as the recommended replacements, pointing to Webflow's vetted app ecosystem for other tools. Any launch discounts attached to that announcement belong to the past, so I am not repeating them.
That is the practical point. A focused identity integration can restore gated content without moving the whole site. If your pages, CMS and editor workflow are fine, the missing piece is login and access, and that is a smaller job than a platform change.
Stripe subscriptions need their own check
For Ecommerce customers with paid subscriptions, the notice said subscriptions stay in Stripe and subscribers continue to be charged. Read that twice. Moving your website does not cancel anyone's subscription, and a new platform does not know those subscriptions exist. Reconcile them separately: who is paying, what access they were promised, and which new system grants it.
One more caution from the notice: Webflow support does not cover third-party integrations, website customisation or custom code. Budget for someone who does.
A validation checklist to run before you commit
This is proposed validation, not a record of work already done. Run it against a staging copy, and treat any step you cannot complete as a reason to pause.
- Inventory current pages and CMS collections, including their relationships.
- Identify forms, protected content and external services (analytics, email, payments, anything calling an API).
- Preserve media, and keep the separately exported CSV content alongside it.
- Map redirects and metadata: titles, descriptions, canonical tags and alt text for every URL you want to keep.
- Recreate one representative content template on the target.
- Test editor preview and publish on that template, using a real editor, not a developer.
- Test forms, access rules and localization.
- Crawl the staging site and compare critical URLs against the live site.
- Rehearse cutover and rollback, including who flips DNS and who decides to revert.
A platform change does not guarantee a search or Core Web Vitals improvement, and nothing here should read as a promise of one. The point of steps 4 and 8 is to avoid losing what you already have.
If you want a second pair of eyes on the inventory, bring the current URL, the requirement that fails and a list of what must keep working to a scoped migration assessment.
The stay-put test
Staying on Webflow deserves a fair hearing. Webflow describes its CMS as visual-first and composable, with draft previews, granular publishing and in-canvas editing, plus reference fields to link content across Collections. Those are vendor claims, but they describe the workflow your editors may already like.
So run the test:
- Does the current editor workflow work for the people who use it daily?
- Does the content model hold your real content without workarounds?
- Is the failing requirement a single integration, such as membership?
If all three answers are yes, the sensible move is to document the smaller change: which integration replaces the failed piece, what URLs and gates it touches, and how you will test it. Replace one integration and leave everything else alone. If the answers are mixed, use the table and the trial tasks. An export is where a rebuild starts, not where it ends.
FAQ
Do I need a paid plan just to leave Webflow?
For code export, yes. Webflow says code export is only available on paid Workspace plans, and a Site plan alone does not include it. CSV exports of collections are described separately in the same guide.
Can I host the exported ZIP and call it a migrated site?
You can host it, and Webflow says no attribution or Webflow plan is needed for it to work on your own server. But collection pages will be empty, forms and search will not work, and protected pages become public. That is a static shell, not a migrated site.
Which is better for editors, Framer or WordPress?
It depends on how your editors work. Framer describes canvas-based editing with preview before publishing; WordPress describes drafts, scheduling, revisions and user roles. Give your own editors the same trial task on both and see which one they can finish without help.
Is Outseta or Memberstack the better pick?
Webflow named both as recommended replacements, and neither source ranks them. Test each against your gate rules, your billing setup and the Stripe subscriptions you already have.
A Webflow export carries markup and assets, not a CMS, checkout, form handling, search, protected pages or extra locales, so choose the route by which of those you cannot afford to rebuild. And if one integration solves the real problem, write that smaller change down before you touch the rest.