When I started InternetDean in 2012, it was me, a laptop, a set of tools, and a handful of local businesses whose networks kept breaking. I had no website. I had no business plan. I had a toolbox and a phone.
Twelve years later we build Cloudflare-powered web applications for clients across three continents. The journey was not a straight line. There were pivots I resisted, projects I should have refused, and a period where I seriously considered going back to full-time repair work. This is what actually happened.
The years that taught the most
- 2012 — 2014 · Fixing things Replacing hard drives, removing malware, reinstalling systems, cabling offices. Unglamorous, and the only genuinely profitable thing I did for years. Almost every client I eventually built websites for first came to me for repairs.
- 2015 — 2017 · The first pivot Clients kept asking whether I could also fix their website. I said yes to a few and learned the hard way that repairing a computer and building a website are different professions. Web work exposed me to design, hosting, accessibility, SEO and project management simultaneously, all at once, badly.
- 2018 · The project I underquoted A booking system for a client who wanted everything, immediately, cheaply. I learned what happens when you accept a fixed price for a fixed scope on work whose scope you did not understand. Delivering it well nearly broke the business.
- 2019 — 2021 · Bilingual work An Arabic-speaking client changed everything. Building a genuinely bilingual site — with proper RTL layout rather than translated text pasted into an English template — turned out to be a technical discipline of its own, and one that opened a market the previous ten years had not touched.
- 2022 — today · Edge-first infrastructure Moving to Cloudflare, Workers and D1. Faster sites, radically less infrastructure to maintain, and deployments that stopped being events.
The mistake that cost the most: saying yes too often
The lesson from 2018 was not "be more selective about clients" as a platitude. It was a specific mechanical change. We now put a written quote in front of every enquiry before any work starts, with scope, deliverables, milestones and price stated plainly.
It feels awkward to send a document that says "this will cost this much and here is exactly what you are getting". What we learned is that the discomfort was never about money — it was about clarity, and it prevented far more difficult conversations later than the one it caused.
Most of the painful parts of running a small studio were avoidable by being specific earlier.
What repair work taught me about building websites
This sounds backwards, but ten years of hardware work is why our websites are fast and our hosting is boring.
When a machine is slow, you do not start by installing more software. You find what is actually consuming the resource. That discipline transfers directly: when a website is slow, the answer is nearly always caching and delivery rather than a bigger server. We measure first, then change one thing.
It also left us with an instinct for durability that a pure-agency background tends to lack. Clients do not need the most beautiful site; they need one that is still working in three years, that they can hand to someone else, and that does not fall over when something unexpected happens on a Sunday.
What we learned from building for Arabic-speaking clients
The bilingual work was the most technically demanding thing we have ever done and the most commercially valuable. Arabic is right-to-left, which means the entire layout has to mirror — not just the text. Getting that right requires CSS logical properties rather than hand-written overrides, separate URLs per language, and reciprocal hreflang tags.
It also taught us that "translated" and "bilingual" are entirely different products. A machine-translated English layout with the text swapped is not a bilingual site. It is an English site that some of the visitors cannot use.
We have written about the technical approach in more detail in our guide to bilingual Arabic/English websites.
The things that are much harder than expected
- Saying no to people you like. Sometimes the right answer is "I am not the right firm for this" and referring them to someone better.
- Maintenance. A launch is a few days. Keeping a site secure, fast and current is the actual job, and it never ends.
- Content. We can build a site that is technically excellent and empty. Clients consistently underestimate how much writing a site needs, and it is the single most common cause of a launch being underwhelming.
- Infrastructure you do not control. We moved a client's site off cheap shared hosting and immediately found a plugin that had been quietly querying their database thousands of times per request for three years.
What twelve years have not changed
Most of our work still comes from someone telling another person. The channels have moved — a WhatsApp message rather than a phone call — but the mechanism has not.
What has changed is the range of problems we get asked to solve. A business that started by asking for network repairs now asks for a booking platform, then a customer portal, then help moving it all to an edge network. Each of those conversations started with someone carrying a laptop into a room.
If you are reading this wondering whether to move your own business online properly, the honest advice is that the first conversation is usually the hardest and the least technical. Get in touch and we will tell you whether we are the right fit — including telling you when we are not.