What we lose, what we gain
This is the real trade of the rebuild, and it should be stated plainly rather than sold. Webflow's visual editor is genuinely excellent — the best free-form design canvas on the market — and moving to a headless CMS gives some of that up permanently. In exchange, everything that is currently impossible becomes a code change. Both halves are true.
Giving upWebflow visual editing
- The free-form canvas. Drag any element anywhere, nudge padding by eye, position absolutely without touching code. Nothing in a headless CMS replaces this.
- Visual CSS editing. The style panel edits any property with instant feedback. After the rebuild, a style change is a CSS change in the repo.
- No-code interactions. Scroll triggers, hover states and timeline animations built by clicking. These become GSAP or CSS written by a developer.
- Building a new page layout unaided. Today a marketer can invent a landing page from nothing. After, they compose from components a developer has built.
- Breakpoint-by-breakpoint visual tuning. Responsive tweaks per breakpoint in the canvas.
- Instant publish. One button, live in seconds. Replaced by a build and deploy — typically one to two minutes.
- In-browser component authoring. Creating a new symbol without a developer.
- One vendor for everything. Hosting, CMS, forms, assets and SEO fields in a single login with a single bill.
GainingDevelopment freedom
- A real Content-Security-Policy. Per-request nonces, no
'unsafe-inline', arbitrary headers and middleware. The reason this project exists, and flatly impossible today. - Git as the source of truth. Branches, pull requests, code review, and rollback to any commit in the history. Every security-relevant change is reviewable and attributable.
- Any package, no platform ceiling. The whole npm ecosystem instead of what one vendor exposes.
- Build-time data. Geocode the events map at build, generate the search index at build. Third-party calls leave the visitor's browser entirely — fewer CSP entries and fewer runtime dependencies.
- Automated quality gates. Header regression tests, Lighthouse budgets, accessibility and visual-regression checks in CI, so a future change cannot silently weaken the policy.
- The WebGL heroes as first-class components. Today they are six unmaintainable copies pasted into Code Embeds with no shared source. After, one component with props.
- Local development and parallel work. Multiple developers on branches without overwriting each other, and no shared staging bottleneck.
- Programmatic content operations. Bulk edits, migrations and audits across 268 records by script rather than by hand.
- No structural ceilings. Collection counts, nesting depth, reference patterns and routing are ours to define.
- Each layer swappable. Astro reads from any CMS and deploys to any host, so no vendor holds the site hostage.
Who actually feels it
The loss is not evenly distributed. Most weekly work is unaffected; the pain is concentrated in one specific activity.
| Task | Today | After the rebuild | Impact |
|---|---|---|---|
| Publish a resource or blog post | Webflow Editor | Sanity, click-to-edit on a live preview | NoneArguably better — real drafts, scheduling and comments. |
| Edit copy on an existing page | Editor mode on the page | Presentation tool, click the text on the preview | NoneSame gesture, same place. |
| Add an event or update taxonomy | CMS collections | Sanity documents with references | None |
| Swap an image or update SEO fields | Asset manager, page settings | Sanity assets, SEO fields on the document | None |
| Reorder or restyle sections on a page | Drag in the canvas | Reorder pre-built blocks in the CMS | ModeratePossible, but only with blocks that exist. New block types need a developer. |
| Build a brand-new landing page layout | Design it in the canvas, publish same day | Compose from existing blocks, or brief a developer | Real lossThe one activity that genuinely gets slower. Mitigated by building a generous block library in Phase 03. |
| Add a scroll animation or hover effect | Interactions panel | Developer writes it | Real lossBetter results, longer path to them. |
| Change a security header or redirect | Enterprise-gated settings panel, or not at all | Edit a file, open a pull request | Large gainThe whole point. |
What comes back, and what does not
A modern headless CMS returns more of the visual experience than the term "headless" suggests. It is worth being precise about the boundary.
| Capability | Comes back? | How |
|---|---|---|
| Edit text where you see it | Yes | Sanity Presentation tool, or Storyblok's visual editor. Click the rendered text, edit in a side panel, see it update. |
| Live preview before publish | Yes | Draft preview on a real URL, shareable with reviewers. |
| Reorder page sections | Yes | A page-builder field of block references, dragged in the CMS. |
| Compose a page from blocks | Yes | Strongest in Storyblok, good in Sanity. Bounded by the block library. |
| Scheduling and draft workflow | Better | Scheduled drafts, comments and tasks — Webflow has no real equivalent. |
| Free-form drag-anywhere canvas | No | Not offered by any headless CMS. This is the permanent loss. |
| Visual CSS property editing | No | Styling lives in the codebase by design — that is what makes the CSP possible. |
| No-code animation authoring | No | Developer-authored. |
The mitigation that decides whether this lands well
Build a generous block library in Phase 03 — hero variants, feature grids, stat rows, quote blocks, CTA bands, logo walls, comparison tables, FAQ accordions — so that "compose a new page from blocks" covers most of what a marketer wants without a developer. A thin library is what makes teams resent a headless migration; a rich one makes the loss theoretical. This is worth over-investing in.
CMS choice
Sanity is the recommendation. Because the editor experience is the biggest risk in this project, the alternatives are priced honestly rather than dismissed.
| Option | Editor experience | Annual, 5 seats | Verdict |
|---|---|---|---|
| Sanity Growth | Presentation tool — click text on a live preview and edit in place. Structured content, not free-form layout. | $900 | Recommended. Strongest Astro integration, 268 documents against a 25K ceiling, free viewer seats, private datasets. |
| Storyblok Growth | Block-based visual page building. The closest thing on the market to composing a page by hand. | $1,188 | The real alternative, and only $288/yr more — five seats included in the flat $99/mo. The stronger pick if editor autonomy is the priority. |
| Payload 3 | Strong admin UI, weaker visual editing. Self-hosted, so you own the database outright. | $0 licence | The most literal answer to total control, paid for in hosting and patching burden. Adds a database to the security surface — a real consideration given the priorities here. |
| Contentful | Enterprise-shaped, weakest visual editing of the four. | from $3,600 | No. Costs more and does less on this site. |
Book a 45-minute demo of both before signing
The stack decision does not depend on which CMS wins — Astro reads from either, and the security posture is identical. Put Chrissy and whoever publishes most often in both editors for twenty minutes each and let them choose. One hour now, and the $288/yr difference stops being the deciding factor.