I Moved My Websites Off of Webflow (thanks to Claude)

I spent the last few months migrating all my websites off Webflow. Not because Webflow is bad, it is actually quite good at what it does, but because the balance of cost and AI powered coding capabilities finally tipped the other direction for me.

A Webflow account plus CMS plans for three sites adds up quickly, especially for projects that aren't monetized and don't need particularly complex implementations. Another Puffy Jacket, my satirical, tongue in cheek outdoor gear site, was never meant to make money. My personal site was just a homepage and a handful of blog posts. The Here & There site had the most elaborate CMS setup, but even that wasn't very wild.

When I first built these sites on Webflow, the value prop made sense. I had more technical expertise than many designers, but not really the time or inclination to manage deployments, databases, and CMS architecture from scratch. Webflow's visual editor compressed all of that complexity into something more approachable. The learning curve, for me, was easy. It gave me the ability to quickly design and tweak without concern for the backend.

Things have changed since then. Tools like Claude Code, Cursor, and Codex have made coding dramatically more accessible. Not in a hand wavy "AI will do everything for you" way, but in a practical way. I can now build, debug, and deploy fairly sophisticated projects without already knowing how to do those things, or at least without needing to know exactly how to do them.

The Rebuild

I rebuilt Another Puffy Jacket, used Supabase for the db, and deployed on Vercel. More importantly, I built supporting tools around it: an Instagram share image generator and a Chrome extension that lets me publish things I find directly to Supabase. These were not things I could have easily done in Webflow, and they make the actual workflow of running the site much simpler. The site costs almost nothing to host, and I have complete control over how it works.

My personal site was simpler. It's a static Next.js build that handles a homepage, blog posts in markdown, and a few case studies. A CMS subscription is overkill for content that rarely changes.

Here & There was a bit more involved because it needs actual content management for ongoing updates. Yes, I know I should probably just migrate my newsletter to Ghost or Beehiiv. I rebuilt it on Astro with KeyStatic as the CMS editor, preserved all the URL structures to avoid breaking links, and ended up with something that is both cheaper and more flexible than the Webflow implementation.

The cost savings are real. Eliminating three CMS plans plus the base Webflow subscription adds up over a year. But the more significant shift is in creative flexibility. Webflow is good at visual website building and teaches you useful concepts about content management and databases. But it is still fundamentally a WYSIWYG system with corresponding constraints. You work within Webflow's model of how websites should be structured and what kinds of interactions are possible in their UX.

Moving to code removes many of those constraints. You work directly with how the stack is configured, how the front-end is set up, how data flows between systems, and how deployment actually functions. The tradeoff is accepting more complexity, and a learning curve that involves more than just clicking and dragging.

The Future

I don't think everyone will abandon visual website builders. Webflow, and other apps like it remain a viable solution for many use cases. But for personal projects where tech saavy people (and increasingly, less tech-savvy) are willing to invest some time learning the underlying systems, the cost/benefit calculation has shifted significantly.

I wouldn't go so far as to say that I'm bearish on Webflow, but they occupy an increasingly awkward position in the market. We've seen the growth of prompt-first website builders like v0 and Lovable (basically no UI, complete abstraction). Webflow sits in a middle ground of power, complexity, and cost. I think there are enough weaknesses that make them vulnerable to attrition from users like me, although the majority use case is more likely to be to a prompt-focused app.

There will be an erosion from people who look at Webflow (and similar platforms) and see something that is too expensive to be worth the (often significant) cost when they could build something themselves with relatively low effort. The gap will likely keep widening as AI tools improve. I would not be surprised to see a similar trend with Figma; I'm spending less time than ever pushing pixels and more shipping production code, fixing bugs, and prototyping locally in our production app.

I still don't consider myself a "developer". But it is far easier today than it was even a year ago for me to build and maintain the sites I need and want.