Moving our own site off Prismic, and what a live site taught us
Our marketing site was built on a setup where every design change needed a developer and a deployment. I rebuilt it in Framer, owning the design and the build. Every existing link survived the move, search rankings held, and page updates went from days to hours. Since launch I've been working through the technical layer underneath, the part that only shows up once real traffic is running through a site.
PROBLEM
The old setup wasn't broken. It did what it was supposed to do. The problem was everything around it.
Updating a section on the homepage meant opening the code, finding the right piece, adjusting it, pushing it to a review branch, waiting for a preview build, getting it approved, merging, and deploying. All of that for a spacing change.
That had a cost beyond speed. When shipping is expensive, you stop experimenting. Ideas get cut before anyone tries them.
SOLUTION
Two things were non-negotiable. Every existing web address had to survive, since we had pages ranking in search, links from other sites, and links shared in emails and posts. And the content couldn't be moved automatically, because the old system stored it in a completely different shape than the new one.
So redirects came first, before any design work. Every old address mapped to its new home, then verified with a crawler after launch to confirm nothing slipped through.
Content moved by hand, deliberately. That meant reviewing every post. Some were outdated enough to retire instead of migrate. Others got updated copy, better images, cleaner descriptions. Slower than automating it, but what ended up in the new system was leaner than what we had.
We also switched to proportional sizing across the whole site instead of fixed pixel values. Small technical decision, outsized effect. Text, spacing, and components now scale together, and adapting a layout for phones became a matter of adjusting one base value instead of overriding dozens.
DELIVERY
A site where design changes go live in the time it takes to select something, adjust it, and publish. Layouts get adapted for different screen sizes visually, as part of designing, rather than as a separate cleanup pass afterward. Where the visual tools run out, we drop into code for the pieces that need it.
IMPACT
The migration went cleanly. Rankings held, no broken link spikes, no lost referral traffic. Page updates went from days to hours, and landing pages that used to take a full sprint now ship in a single focused session. The real win is operational: the distance between "what if we tried this" and "it's live" is measured in minutes.
RETROSPECTIVE
I'd inspect the raw output earlier. I ran the automated audits first because they're fast, and I believed a false result for longer than I'd like to admit before I opened the page source and saw the tool was wrong. The manual check took an afternoon. It should have been step one.
I'd also treat the launch checklist as half the job, not the whole thing. Redirects and content got obsessive attention. Social previews, per page descriptions, and the contact form got "we'll do it after." After turned out to be a long time.
MORE PROJECTS
Running a fast growing SaaS site solo
Three languages, a growing product line, 175+ posts, one person keeping it all coherent. I extended Pentagram's design system with everything that came after it. Active users doubled, the site never slowed anyone down.
A Fortune 500 team was bouncing between seven tools to get one piece of data. I led the design of the portal that replaced them all. Requests dropped from 12 days to 30 minutes.