Keeping the editor, losing the renderer
There is a version of leaving Webflow that does not take the CMS away from the people who use it every day. You keep an editor they can log into, and you move the rendering into your own codebase. The split is real. So is the part it does not fix: the content still lives in a database you rent.
What the split actually buys
The renderer is where the performance problems live. Webflow decides how a page gets assembled, which scripts load and in what order. When Stay World Class moved off Webflow and Xano onto Next.js, NestJS and Supabase, their Lighthouse score went from 55.91 to 91 out of 100 and LCP got 77% faster. None of that needed a different CMS. It needed control over the HTML.
Once you own the rendering, ordinary engineering comes back. You can write a test for the page. You can cache what you want, ship a component library, and host the site at a price that has nothing to do with how many editors you employ. Meanwhile the marketing team logs into the same admin as yesterday. That is most of the win for a much smaller fight than a full rebuild.
The bill moves, it does not disappear
Sanity's Growth plan is $15 per seat per month, and the metering underneath it is the part worth reading. The plan includes 25,000 documents; going past that is an add-on at $299 a month for 50,000 in total. API requests are capped at 250,000 a month, then $1 per 25,000. A third dataset beyond the two included costs $999 per dataset per month. Bandwidth past 100GB runs $0.30 a gigabyte.
Contentful meters a different way. Its included Starter Space holds 10,000 records, 25 content types and 2 environments. The upgrade is a Lite Space at $850 a month, and Contentful's own table notes you are limited to one of them. The free tier allows 100,000 API calls a month with no overages, which means it stops serving rather than bills you.
Neither of those is unfair pricing, and I would not argue otherwise. It is the same shape as Webflow's, though. The number grows with your content and your traffic, not with your engineering. So if the reason you are leaving is a bill that scales against you, a hosted headless CMS moves the bill without changing its direction.
Owning the database is a separate decision
If you want the content in a database you actually hold, pick a CMS that runs on your own Postgres. Payload is open source and self hosted, and its official adapters cover Postgres and MongoDB, with SQLite for smaller deployments. Content then sits in tables you can query and back up. The Stay World Class migration ran a four stage ETL that mapped legacy IDs onto PostgreSQL UUIDs so no record lost its history, and you can only write that ETL when you can reach both ends of it.
Self hosting does not make you free of other people's decisions. Payload joined Figma, and Payload Cloud has paused deployment of new projects while existing ones keep running. The open source core is unaffected and the company says current Cloud customers will be helped to move when the time comes. Still, anyone who picked the managed option is now planning a migration they did not schedule. Running the CMS on your own infrastructure is what turns that announcement into a shrug.
When Webflow is still the right answer
A twelve page marketing site with one person editing it does not have a renderer problem. Webflow's Premium site plan is $25 a month billed yearly and includes the CMS. At that price, the engineering hours to replace it cost more than a decade of subscription, and you would be buying control over a page nobody has complained about. If your content team ships weekly and nobody has filed a performance ticket, stay where you are.
The number that should make you do the arithmetic is $2,500 a month, which is what Webflow now lists for its Team platform plan, annual contract required. If you are climbing to that tier for publishing workflows and governance rather than for traffic, you are paying platform prices for process. That is the point where rendering your own pages and keeping an editor your team already likes stops being a preference.
