The site you’re reading this on just went through a full redesign. Same stack underneath — Astro, hand-written CSS, no React, no build-pipeline surprises — but a genuinely different feel on top. Here’s what that actually took, including the unglamorous parts nobody puts in a portfolio case study.
Starting with three questions, not a mood board
Before any layout work started, we forced ourselves to answer three questions plainly: who is this actually for, what problem does FluxaraX solve for them, and what’s the promise. Skipping straight to “make it look premium” is how you end up with a generic AI-generated site — gradient text, purple everywhere, glowing orbs, the same card grid as every other portfolio site built this year.
The answer, kept deliberately unglamorous: Australian small business owners who need a working website, app, or AI tool, don’t have an in-house dev team, and are sick of agencies selling proposals instead of shipping. That’s who every design decision downstream had to serve — not a hypebeast audience, a 40-something tradie or real estate principal.
The hero: a forest that opens into a sunrise
The concept was simple to describe and harder to build: a dark, monochrome forest path that parallax-parts as you scroll, revealing a photoreal orange sunrise where the path leads. Dark and enclosed opening into light — a literal version of the “new opportunities” idea in the site’s actual promise, not just a nice visual.
The technical part is layering — background, midground, and foreground moving at different scroll speeds, which is the single biggest lever for making a scroll effect feel like real depth instead of a flat parallax gimmick. The sunrise itself was generated, the layering and scroll-binding was built by hand. Getting the parting effect to feel natural rather than mechanical took several passes.
The part that doesn’t show up in a screenshot: verification
The build used a screenshot-verification loop as a hard rule, not an optional step — every scroll effect, every animation, gets screenshotted frame by frame and checked before moving on, including a separate explicit pass for mobile. Desktop and mobile scroll behaviour are not the same thing, and assuming they are is how sites end up broken on the one device most visitors actually use.
The portfolio card stack
The old portfolio section was a static grid. The new one is a proper card stack — drag and swipe, keyboard navigation, spring physics on the fan effect, consistent hover behaviour across every card. Small detail, but it’s the kind of thing that either feels genuinely responsive under your thumb or feels janky, and there’s no real middle ground once you’re paying attention to it.
Fixing the thing that actually mattered most
Buried in the middle of all this: the contact form had been sitting on a dead placeholder endpoint. Nobody who filled it out was reaching anyone. Redesigning a hero is the fun part. Confirming a form submission actually lands in an inbox — sending a real test, checking the actual email arrived, not just that the form submitted without an error — is the unglamorous part that matters more. Both got done, in that order of priority even if not that order of visibility.
Merging without breaking anything
The last step wasn’t the redesign itself — it was making sure production matched what was verified in preview. Every page checked for a clean 200 response, every image asset confirmed loading, the contact form’s real endpoint confirmed present in the actual production HTML, not just the preview build. Preview and production can drift on things like environment variables in ways that don’t show up until someone’s actually trying to use the live site.
What’s next
Structured data’s gone in since, a sitemap’s live, and we’re actually publishing to this blog again instead of leaving it at one post from April. Slow and steady beats a redesign that looks great and then goes quiet for five months.