You work with me, directly

In August a client messaged me at 2 a.m. Card payments on their site had stopped working: a Stripe API key had failed. I had payments running again within thirty minutes, then changed the code so that key cannot take them down the same way again.

The fix was the easy part. The hard part was that the site gave no warning. They found out because a customer told them. A broken payment does not send an error. It just goes quiet, and every hour it stays quiet costs money nobody sees leaving. That night is the clearest picture I have of what I actually sell: not the site, the part behind it that has to keep working when nobody is looking.

I started in graphic design: banners, logos, the odd client. One evening, working on a website concept for an online bank, I spent ten minutes looking for a name and ORD came to me. It first stood for Original Responsive Design. The work outgrew design; the name stayed.

My first web client came in February 2026, a hair studio in London taking bookings through Instagram and WhatsApp messages, with no booking system behind them. I rebuilt the site, then in May moved it to a new domain. It was my first migration, and the new domain started from zero. After that I built the blog's content system and published more than twenty articles. By 22 September 2026, Search Console showed 157,000 impressions and 2,120 clicks.

I improve the system behind a business: the website, the client journey, booking, intake, follow-up, automation and the backend that holds it together. I find where you are losing time, leads or revenue, and I fix that part.

The website is one component of that, and it is usually not the most expensive thing that is broken. A site can be beautiful and still sit on top of a process that loses a share of everything it receives. So I look at your site the same way I look at your CRM and your follow-up: one connected system that either supports revenue or quietly loses it.

Every engagement starts with an audit, delivered in 5 working days, and a written plan two days after it. That order is not a formality. Scoping work before looking at the system is guessing with a price attached, and it is how businesses end up paying for a redesign that does not move anything.

After that, scope and price are fixed and agreed before any build starts. You own the code, the accounts and the documentation the whole way through. Nothing is locked to me, and if you decide in a year to hand it to somebody else, they will find a codebase with comments explaining why things are the way they are.

I take on a limited number of projects at a time. That is a real constraint rather than a positioning line: it is the reason the person who scopes your work is the person who does it, and the reason I would rather turn something down than take it and be slow.

Next.js, n8n, Notion and Cal.com. It is a deliberately small list. I would rather know a handful of tools deeply enough to know where each one breaks. The strip above shows the tools that have appeared in projects; these four are the ones I rely on every week.

The choice between them is made on what your setup costs to run, not on what I enjoy building in. If you already pay for something that does the job, that is what gets used. Adding a platform to solve a problem your existing platform could solve is a cost you carry every month forever. Payments are the clearest case: if you take them through Stripe, I work inside your Stripe account and connect it to your site, rather than bringing a payment setup of my own.

Every business in this field claims real results and almost none of them show a figure with a name next to it. So the before and after numbers go on the site, attached to a named client, where you can push back on them. There is exactly one case study up at the moment, which is Matted London, and it carries a Search Console figure, not only a testimonial about how smooth the process was.

The same rule applies in reverse. If a number cannot be traced to a project with a name on it, it does not go on the site. An earlier version of the automation page claimed a weekly hours-saved figure with no client behind it, and it has been removed rather than softened.

I work remotely, in English, with clients in the UK and in Moldova. Most conversations start with a call or with an emailed URL, and either is fine. If you would rather not talk to a stranger before you know whether this is worth your time, send the URL and you will get a written teardown back within two working days.

That last option exists because I have been on the other side of it. Being asked to give up half an hour before finding out whether someone can help is a real cost, and most of the field asks for it anyway. If the teardown tells you the answer is no, that is a good outcome and it cost you one email.