How long it takes to build a website / Captain’s log, entry 32

Stardate: 26169.9

A quick note for context: if you’re B2B, like we are, a website earns its keep when it works as a clean presentation point and a reliable lead-gen hub.
I got to that the long way round:

1. I started with social accounts.
2. Then I shipped the first site on Tilda before New Year 2022.
3. Then life happened…
4. Then it got outdated fast: the team changed, the concept evolved, and I buried that site.
5. Then I realised I needed a way to present us and our capabilities, so I built a Google Slides deck for that.
6. It went through 5 revisions, and I realised I NEEDED A WEBSITE, because it’s simply convenient: you change something on the site, and everyone you’ve shared the link with sees the update. Before that, one person had version 3 of the deck, someone else had version 5, someone still had the very first one…
7. I spent the next 8 months building the site through ChatGPT 4o dialogues. Codex for code work appeared along the way, but back then I didn’t get that it was meant for me, so I didn’t learn it. I’d have finished much faster.
8. Then it clicked that I should be aiming at the European market. More broadly, at an English-speaking market. That’s part of the whole point of leaving Russia: becoming diversified, untied from the Russian economy, which doesn’t look like it’s headed anywhere good. That’s why the site redesign was already on the table by late summer 2025. The idea of plugging myself into non-ruble economies showed up for me back in November–December 2022, maybe even earlier.
9. Then I realised it would be easier to keep two versions of the site — Russian and English — on separate servers, with different Privacy Policies and different compliance procedures. So I decided the site should move fully onto English rails.
10. I did research on The Buyer’s Persona with the goal of getting European clients.
11. I realised we’re basically nobody to them, and that’s where the “Hungry Team” positioning turned out to be useful.
12. I wrote a plan/brief for translating the site into English and reshaping it around the experienced European buyer persona + making it clear that we’re a hungry team.
13. I figured out what Codex from OpenAI actually is, and I realised it can speed me up by ~10x while raising the quality of solutions. When you build code via GPT chat, you end up with a familiar routine — and to get anything finished, you have to:
    — feed it a sh**load of context first (copy the codebase of 3 files for a single page: php, css, js),
    — keep the whole concept and each page’s job in your head (Codex has the same need, but there you can store it in a .md file and let Codex update it; the catch is that the file goes stale fast if you don’t keep it current),
    — ask it to POINTOUT what’s broken, where it’s broken, and ask it to give code in a PRECISE AND SURGICAL way — so you can paste it into the WP theme editor and see what actually worked.

Anyway, Codex + GitHub version control is a different universe in terms of effectiveness.
I learned git in full monkey-mode: pull, switch, merge, push and….

14. Translating the site into English and finishing the UX/UI under The Buyer’s Persona — through the Codex + GitHub workflow, local testing on my laptop, then shipping the finished build to hosting via SSH — took…

129 hours 43 minutes, which is:
— if you work 8 hours a day on this one task: 16 days, 1 hour, 43 minutes,
— if you live without sleep and work 24 hours a day: 5 days, 9 hours, 43 minutes

In reality, some days I worked on it all day, some days half a day, and the most active phase took about a month and a half.
Total time was ~2.5 months, because I created the Codex repo back in mid-December.