Decisions — calls I'd make again
Decisions, and why.
How I decide, the calls I made, and the ones made under pressure: a deadline that couldn't move, traffic peaks, and money on the line.
( 01 ) — Principles
How I decide.
- P/01
Measure before deciding.
Start from the numbers, find the causes, then fix them.
From: From about 55% to 95%+ on Lighthouse→
- P/02
Write the trade-offs down.
A written comparison persuades better than an opinion, especially outside frontend.
From: Arguing for Next.js, in writing→
- P/03
Reuse when the deadline is fixed.
Less to build means less to test, and a launch that happens on time.
From: Building the World Cup booking inside the platform→
- P/04
Protect the money first.
Payments and wallets get tested before anything else, on both sides of the API.
From: Test the paths that move money first→
( 02 ) — The calls · 6 decisions
The calls I'd make again.
Build the platform on Next.js
Instead of: HTML, Sass and Blade
Scalability, maintainability, development speed, and product pages that render with their own metadata for SEO and link previews.
What happened. The comparison supported the team's technology decision. Product pages became shareable and indexable, and seven live storefronts run on the platform today.
Put the money paths at the top of the test plan
Instead of: Spreading testing evenly across every screen
Payments and the wallet are where a bug costs merchants and customers real money.
What happened. Releases were checked where a mistake would cost the most.
Integrate World Cup booking into the existing platform
Instead of: Launching a separate website
It reduced the implementation scope, so there was less to build and test before a deadline that could not move.
What happened. Less to build and test under a very tight timeline, and the experience supported about $20,000 in revenue per match event.
Check Lighthouse before every merge
Instead of: Auditing after launch
Scores drift as features land. Catching a regression in review is cheaper than finding it in production.
What happened. The storefronts stayed at 95%+ across performance, accessibility, SEO, best practices and PWA readiness.
Cut steps out of the booking flow
Instead of: Keeping one screen per field
Each extra screen between discovery and reservation is a place to drop off.
What happened. A more direct path from discovery to reservation.
Treat designs as a starting point
Instead of: Copying every pixel as a fixed spec
Implementation reveals UX problems a static screen can't show.
What happened. Responsive, production-ready interfaces built from the designs, improved where it mattered.
( 03 ) — Under pressure
When it got hard.
Decisions made with the clock running, traffic spiking or money on the line.
Konin Cloud
One click, one payment
The pressure: A payment flow, where a duplicate request could mean a duplicate payment.
I disabled the button while the request ran and agreed a duplicate-request check with the backend team, so the protection didn't depend on the UI alone.
Outcome. There were no duplicate requests in testing.
Full story →Konin Cloud
The bug that only Safari saw
The pressure: Caught right before a release, fixed before it shipped.
Cross-browser checks before the release caught it, and I fixed it with a fallback before shipping.
Outcome. Users never saw it.
Full story →The Sailors
Building the World Cup booking inside the platform
The pressure: A fixed, very tight World Cup timeline.
I proposed integrating it into the company's existing platform rather than launching a separate website, then built and launched it with React, TypeScript and Material UI.
Outcome. Less to build and test under a very tight timeline, and the experience supported about $20,000 in revenue per match event.
Full story →The Sailors
When match tickets opened, traffic spiked
The pressure: Traffic peaks at every ticket release.
I kept the booking path minimal, cached match information and lazy-loaded the heavy parts.
Outcome. Booking stayed usable during the peaks.
Full story →
Every story, one place
Stories→