Start here · UK · 2026
How to start a web design agency with no experience
Five years ago, starting a web design agency meant one of two things: learn to code for a year, or subcontract the work and keep pennies on every job. Both barriers are gone, and hardly anyone has adjusted to what that means. An AI coding agent will now take a plain-English instruction and return a complete, mobile-first website in hours, which sounds like it should have destroyed the web design business. It did the opposite. It destroyed the price of the build and left the actual business of finding clients, winning trust and looking after sites month after month wide open to anyone willing to do it. Roughly £25 a month of tooling now covers what an employed developer used to do. The people winning with this aren't designers or coders. They're the ones who picked a patch, made the calls, and sold an ongoing service instead of a one-off build. Web design is the easiest way in, though it is really one service inside a wider AI services agency. Here is the whole route, honestly, including the parts that are still hard.
To start a web design agency with no experience in 2026, you don't learn to code, you run a process: pick a local niche, use an AI coding agent such as Claude Code to build demo sites in hours, and sell a done-for-you plan at £50–150 a month instead of one-off builds. The tooling costs about £25/month. The hard part isn't the websites; it's consistent outreach.
Sell retainers, not websites
A new web design agency in 2026 should sell a done-for-you monthly plan at £50–150 a month, with hosting, security and edits included, because AI has collapsed the price of the build itself while the value of ongoing care has gone up.
This is the single decision that separates an agency from a string of odd jobs, and it's worth getting right before you build anything.
Why the one-off build is a dead end
When a website took eighty hours to hand-build, charging £1,500–£3,000 once made sense. Now that the build takes hours, one-off pricing puts you in a race to the bottom against every DIY builder advert on television. Worse, a one-off client pays you once and disappears, so every month starts back at zero. The market has split: the cheap one-off end belongs to Wix and Squarespace now, and the end that's growing is done-for-you. These are businesses that don't want a login, a dashboard or a tutorial. They want it handled.
What the monthly plan includes
The plan that sells is simple: I build it, I host it, I keep it working, and small changes are included, from £50 a month for a one-page site and £100–150 for a full site. Hosting at this scale costs you pennies, so almost all of that is margin for the ongoing relationship. Ten clients at £100 a month is £12,000 a year of recurring revenue, and each new client stacks on top of the last instead of replacing it.
| One-off build | Monthly plan | |
|---|---|---|
| The client pays | £1,500+ once | £50–150/month |
| Their decision feels like | A big invoice to approve | A line in the direct debits |
| Your income at month 12 | Resets to zero each build | Compounds with every client |
| After go-live | You're gone; site slowly dates | You host, edit and retain |
| Who you compete with | Wix adverts and DIY builders | Almost nobody doing done-for-you |
| Natural upsells | Rare, no relationship | SEO pages, blogs, reviews |
Why small businesses say yes
A £1,500 invoice is a board meeting for a sole trader; £50 a month is a line in the direct debits next to their van insurance. It's also directly comparable to what many of them already pay lead platforms, often £60–120 a month for shared leads they don't own. A flat monthly fee for an asset with their own name on it is an easy comparison to win.
Same ten clients, three pricing models. One-off income stops; retainers keep collecting.
The toolchain: what replaces the skills you don't have
The working stack for a no-experience web design agency is an AI coding agent (Claude Code), a free GitHub account and free static hosting such as Cloudflare Pages: roughly £25 a month all-in, most of which is the AI subscription.
There's no web design software to master and no course in HTML required. The stack is three accounts and an afternoon of setup.
What the AI genuinely does
Given a clear instruction (the trade, the town, the photos, the phone number), a coding agent writes the entire site: layout, copy, mobile behaviour, contact forms, the lot. It will rebuild a prospect's dated website onto a modern template in an evening. It handles the technical plumbing (domains, deployment, forms) with you directing in plain English. That's not a gimmick; it's the same class of tool professional developers now use daily.
What it doesn't do: the honest bit
The AI will not find you prospects you haven't pointed it at, pick up the phone, or follow up politely two weeks after a "maybe". It also won't stop you shipping something generic: you still make the calls on taste, meaning which photos, which testimonials, what the first screen says. Anyone promising the whole business runs itself is selling you something. The build is automated; the business isn't.
What it costs to run
Claude's paid plans start at roughly £15–18 a month (current rates are on Anthropic's pricing page), domains are about £10 a year each, and GitHub plus Cloudflare Pages hosting are free at this scale. Call it £25–50 a month to run the whole operation, less than most people's petrol for a week. The startup cost is genuinely the lowest of any service business I can think of.
| Tool | What it does | Cost |
|---|---|---|
| Claude (paid plan) | Builds the sites, the entire technical skill set | ~£15–18/mo |
| Domain per client | Their address on the web (often billed to the client) | ~£10/yr |
| GitHub | Stores every site's code, keeps history | Free |
| Cloudflare Pages | Hosting, SSL and global speed | Free at this scale |
| Total to operate | The whole agency, before your first client | ~£25/mo |
Pick a niche and a patch before you build anything
The easiest first market is local trade businesses (plumbers, electricians, decorators) inside a defined territory, qualified by two signals: a visibly dated website and a mobile number on it.
"Everyone with a business" is not a niche, and a beginner competing everywhere competes nowhere. Constraint is what makes the outreach workable.
Why local trades
Trades are the ideal first niche for a reason: demand for their work is local and constant, an enormous number of them have websites built in 2012 or none at all, and they're used to paying monthly, often resentfully, for marketing on lead platforms. You're not educating them about why being findable matters; you're offering a better deal on something they already buy. If you're weighing this against flashier options, why local trades beat SaaS, startups and restaurants as a first niche makes the full case.
The two-signal qualification bar
A dated website means there's a visible, undeniable problem you can show them. A mobile number on that site means you can actually reach the decision-maker: no gatekeeper, no info@ inbox where messages go to die. Every prospect must clear both bars before they earn a place on your calling sheet. This one rule filters out most of the wasted effort beginners burn on prospects who were never reachable.
How big a territory
A county, or a city and its surrounding towns, is plenty: enough to hold hundreds of qualified prospects across five or six trades, small enough that "I'm based near you" is true on every call. Local familiarity closes deals: naming their town, knowing their patch, being someone they could theoretically meet.
Get your first client with a demo, not a portfolio
You don't need a portfolio to win a first client. You need one demo: a real prospect's existing website rebuilt onto a modern template and sent to them as a live link before you've ever spoken.
The chicken-and-egg problem (no clients without a portfolio, no portfolio without clients) dissolves when the build costs you an evening instead of a month.
The swap-and-send demo
Take a qualified prospect from your sheet and rebuild their actual business (their trade, their town, their photos) as a real, scrollable site on your template. Then send them the link. A screenshot is forgettable and a pitch is ignorable, but their own business looking better than it's ever looked, live at a link on their phone, starts a different conversation: not "would you like a website?" but "this already exists, want it?" Each demo you make also becomes portfolio for the next call.
The call and the follow-up
The demo earns you the right to a short, honest call: you saw their site, you built them something better, it's £X a month done-for-you, no setup fee. Most won't say yes on the first contact. The money is in polite, spaced follow-up over the following weeks, because you're often just waiting to be there in the week their old site finally embarrasses them in front of a customer. The full loop, from lead sheet to booked call, is broken down step by step in how to get your first web design client.
The legal bit nobody teaches
UK business-to-business cold outreach is legal done properly: screen numbers against the Telephone Preference Service and Corporate TPS before calling, follow PECR rules for emails and texts, and handle prospect data under UK GDPR. Gov.uk's direct marketing guidance is the plain-English starting point. Sole traders count as individuals under these rules, so screening isn't optional. None of it is difficult; it just has to become routine before the first call, not after the first complaint.
The boring setup and the realistic timeline
In the UK you can start as a sole trader with a single registration on gov.uk, and a realistic timeline from blank laptop to first paying client is four to eight weeks of consistent outreach.
The administrative side is smaller than most people fear, and the timeline is longer than most courses promise. Both facts matter.
Registering and getting paid
Setting up as a sole trader is a free HMRC registration, with no company formation needed on day one. For payment, monthly direct debit (GoCardless and similar) beats chasing invoices: the client signs once and the retainer collects itself. A one-page agreement covering what's included, what's extra and how either side leaves is enough to start.
A week-by-week shape
Weeks one to two: set up the tools, build your own agency site (it's the answer to "have you done this before?") and your demo template. Weeks two to four: build the lead sheet for your territory and start daily calls with demos going out ahead of them. Weeks four to eight: follow-up season. This is where the first yes usually lands, and where most people who fail have already quietly stopped calling.
| Phase | Weeks | What actually happens |
|---|---|---|
| Setup | 1–2 | Tools installed, your own agency site live, demo template built |
| Prospecting | 2–4 | Lead sheet filled for your territory; daily calls with demos going out ahead |
| Follow-up | 4–8 | Where the first yes usually lands, and where most people quit |
Who shouldn't start an agency
Skip it if you won't do outreach in any form, if you need income this month rather than this quarter, or if you're shopping for passive income. This is a real service business with real customers, and it compounds precisely because most people won't do the unglamorous fortnight of follow-up calls. If that filter doesn't put you off, the barrier that used to be technical simply isn't there any more.
Everything above is the compressed version of a system that runs today at a real UK agency: sitework.uk, which sells exactly these £50–150/month plans to local trades. The full system (the prompt library, the lead-sheet method, the call scripts, the legal module and one client followed end to end) is documented as a 14-module course, and module 00 is free, so you can judge the method before any money moves.