Blennex
I built this site twice. The second time was my own doing.
- Role
- Development partner, and social media
- Client
- Blennex, design agency, Pakistan
- Built
- React and Sanity CMS · 18 July 2025
- Rebuilt
- Next.js, TypeScript, Tailwind · 5 June 2026
The first build
Blennex is a graphic design agency based in Pakistan and working internationally, specialising in packaging and label design and 3D mockups. I am their development partner: they design, I build, and I look after their social media.
Their own site was the first thing I made for them, in July 2025, in React with Sanity as the CMS. On paper that was a reasonable call. They needed to change their own work without asking me, and a hosted CMS gives you that on the first afternoon.
What “slow” actually meant
The site felt slow, and the CMS was the reason. Every page had to ask for its content before it had anything to show, and that request sat on the critical path rather than beside it.
On most sites you can absorb that. Not on this one. A design studio is judged on how its work looks, and nobody reaches the work until the page has finished fetching it. The delay landed exactly where it did the most damage.
Two other things came with the same decision. A crawler asking for a page received a shell and had to execute JavaScript before there was anything to read, which is a poor trade for a site whose value is entirely in its content. And the contact flow, the one element on an agency site with a job to do, was not doing it.
Optimise or rebuild
Worth being clear that this was a real decision and not an excuse to start again. Optimisation is the right answer far more often than developers like to admit, and a rebuild is the expensive option that feels like progress.
It was the wrong answer here because the cost was not in any one component. It came from when the page assembled itself. I could have spent weeks trimming bundles and caching responses and would have been optimising toward a ceiling set by the architecture. So I moved the ceiling.
The rebuild
In June 2026 I rebuilt it in Next.js with TypeScript and Tailwind, deployed on Vercel. Pages are rendered ahead of the request, so they arrive as content instead of as instructions for assembling content. That answers the speed problem and the crawler problem with the same change, because they were always the same problem wearing two hats.
I reworked the contact flow while I was in there, and fixed the design and layout issues that had built up in the first version rather than porting them across intact.
What I can and cannot claim
The mobile scores above are the measured part, and accessibility is 91 rather than 100, so this is not a clean sweep. There is no business number to sit beside them yet — no lift in enquiries I could stand behind. The site is recent. When there is one, it goes here.
Since launch
I still maintain the site and ship new features, alongside the client work I build for them.