Why I rebuilt a Webflow site in React

I built the original site in Webflow. It was responsive, fast and easy to update. For a while, it did exactly what an actress portfolio needed to do.

The rebuild did not start because Webflow suddenly became bad. It started because the useful decisions were accumulating at the edges of the template. The reel needed to play before someone chose to click it. Measurements needed to make sense in different countries. A filmmaking page needed to behave differently from a list of acting credits. Each exception was small. Together, they had become the site.

The first exception was the showreel

The Webflow version showed a poster image with a play button. That is a perfectly normal pattern, but it put the first useful piece of evidence behind a speculative click.

A casting director may be moving through a large stack of portfolios. They can judge lighting, framing and performance quality from the first few seconds of footage, so the rebuilt reel starts silently and immediately. The visitor is already watching before they have to decide whether watching is worth it.

The change was not technically impressive. That was the point. In React, the behaviour lived with the component instead of in a separate script that had to be remembered later.

Then the content stopped fitting the template

The next update arrived as more than 500 lines of credits, training and representation details. The structure had changed too. There were new categories, different agents in different countries and measurements that needed both metric and imperial units.

I used Claude to map the document into the existing content files, then changed the components around the material. The regional data led to a measurement switcher. The training history needed a clearer hierarchy. Contact details became a small representation system rather than one generic email button.

That workflow changed the order of the work. I did not have to force new content into old fields before I could see the design problem. The content and the interface could change together.

The switch was not really about React

A visual builder is still a good way to launch a site whose structure is understood. It became less useful here once the exceptions carried most of the meaning. At that point, changing the behaviour, data and layout in one place was simpler than maintaining a growing layer of workarounds.

React and AI did not make the design decisions. They made it cheaper to test and keep them. Webflow helped me get the first version live. Code made the later versions easier to reason about.