You Built A Site For A Mate. Now You Sell Websites. Congratulations, You Are A Hazard.

There is a specific LinkedIn post I see roughly nine times a week now. It goes something like this: 'Built my mate's barbershop website over the weekend. He loves it. Taking on 3 clients this month at £1,500 a pop. DM me if you want in on the founding rate.' There is always a screenshot. The screenshot is always the same Framer template with the images swapped and the client's name in Bebas Neue. The comments are always the same: 'Amazing! So inspiring! DM'd!' And somewhere in the middle of all this, a small business owner is about to hand over the digital shopfront that pays their rent, feeds their kids, and covers their van finance to someone who has, in the strictest possible sense, built exactly one website.

I have been doing this for seventeen years across roughly two hundred sites. I want to be gentle here because the enthusiasm is genuine and the ambition is admirable. But I also want to be honest, because the person the soloprenuer is about to invoice is not another soloprenuer. It is a plumber. A dog groomer. A physiotherapist. A woman who bakes wedding cakes out of a converted garage. These people are not buying a website. They are buying the mechanism by which strangers will find them and give them money. If that mechanism doesn't work, the mortgage doesn't get paid. That is the actual, non-abstract stake. And the weekend Framer flip does not begin to reckon with it.

The iceberg problem

Web design and development is an iceberg. The bit above the waterline — the pretty design, the fonts, the hero image of a smiling person holding a clipboard — is maybe 10% of the actual job. It's the part clients can see, so it's the part they think they're buying, and it's the part the beginner mistakes for the whole discipline. Underneath the waterline is the 90% that nobody photographs for LinkedIn. It is the boring, unsexy, business-critical scaffolding that determines whether the site earns its keep or quietly haemorrhages the client's savings for two years before they give up and open a Facebook page.

Here is a non-exhaustive tour of what lives below the waterline. Skim it and ask yourself, honestly, how many of these you'd know how to do on the barbershop site next Tuesday.

• →DNS, MX records, SPF, DKIM, DMARC — because when the client's Gmail stops working the day after launch, that's now your problem, and 'have you tried turning it off and on again' is not a fix.
• →SSL, HSTS, certificate renewal, mixed-content warnings, and the specific joy of explaining to a client why Chrome is telling their customers the site is 'not secure' six months in.
• →Backups. Actual backups. Off-site, versioned, tested restores. Not 'my hosting has backups' — that is a hope, not a plan. When the site goes down at 4pm on the Friday before a bank holiday, you will find out whether you have a real backup or a comforting lie.
• →GDPR. The UK-GDPR. The ePrivacy directive. Cookie consent that actually blocks cookies before consent. A privacy policy that reflects what the site actually does, not a copy-pasted template from a US law firm mentioning 'California residents.'
• →Accessibility. WCAG 2.2 AA is now a legal floor in the EU under the European Accessibility Act as of June 2025. If the client sells to Europe and a screen reader can't complete their booking form, you have exposed them to a claim. Your Framer template did not test that.
• →Performance. Core Web Vitals. LCP under 2.5s on a mid-tier Android in a Cornish car park with two bars of 4G. Not on your MacBook Pro on fibre.
• →SEO that isn't 'I installed Yoast.' Schema markup, canonicals, hreflang if they trade across borders, an XML sitemap that isn't fighting the CMS, redirects from the old site so they don't lose the rankings they spent a decade building.
• →Analytics that respect consent, funnels that actually measure the thing the business cares about, and the wisdom to know that 'sessions up 40%' means nothing if enquiries are flat.
• →Forms that reach an inbox someone reads, with spam protection that doesn't block legitimate customers, with autoresponders that don't land in Promotions, with a fallback for when SMTP dies at the ISP.
• →Hosting that matches the workload. Not the cheapest shared plan. Not the most expensive managed platform because 'it sounds professional.' The one that fits.
• →A staging environment. Because you do not push edits directly to the live site of a business that takes £4,000 bookings.
• →Version control. Because 'index-final-FINAL-v3.html' is not a deployment strategy.
• →Uptime monitoring. Error monitoring. Log retention. A plan for the 3am outage that doesn't involve the client discovering the outage before you do.
• →Ongoing maintenance. Security patches. Plugin updates. PHP version bumps. Framework migrations. The Google algorithm changing its mind about what a good page looks like on a Tuesday.
• →A contract. Actual T&Cs. A limitation of liability. Professional indemnity insurance. Because if the site takes payments and something goes wrong, 'we were mates' is not a legal defence.

That is not a comprehensive list. It is the visible tip of the underwater portion. Underneath that are conversations about brand positioning, conversion psychology, tax implications of the checkout flow, PCI-DSS scope if you touch card data, and the small matter of whether the domain is registered to the client or to your personal Namecheap account — because if it's the latter and the relationship sours, you have accidentally taken a small business hostage.

Why the barbershop site 'worked'

The mate's site worked because the mate was already going to give you the benefit of the doubt. The barber has an existing customer base who find him by walking past the shop. The website is decorative. It is a business card that lives on the internet. If it goes down for a fortnight, three people notice, and two of them are his mum. This is a completely different job to building the online engine for a business that has no walk-in trade and no other channel. Those two jobs share a URL structure and almost nothing else.

“The client is not paying for a website. They are paying for the outcomes a website makes possible. If you don't know what those outcomes are, you are not selling a service. You are selling a Figma export with an invoice attached.”

The reason experienced agencies charge what they charge is not because we've all conspired to keep prices high. It's because a functioning small-business website is a nine-month conversation, a two-year maintenance commitment, and about £4,000 in tooling and infrastructure a year before we've written a line of code. It is professional indemnity insurance. It is a business bank account and a proper limited company because when things go wrong you do not want the client suing your personal savings. It is the third phone call at 8pm on a Saturday explaining to a florist why her Mother's Day landing page is 404ing and what you're doing about it right now.

The market is about to sort this out for you

Here is the awkward bit. I am writing this at the exact moment the barrier to producing a passable-looking website is collapsing to roughly the price of a coffee. Lovable, v0, Bolt, Cursor, the Framer template farms, the Webflow flippers, the WordPress-block-theme brigade. Anyone with a laptop and an afternoon can produce something that looks, to an untrained eye, indistinguishable from a £5,000 agency build. That's real. That's the ground genuinely shifting. But the visible output converging does not mean the underlying job converged. It means the top 10% got easier and the bottom 90% got harder, because there are now more sites in the wild being run by people who do not know the bottom 90% exists.

The market will sort this out. It always does. It sorts it out slowly, in a way that is quietly ruinous for the clients who happen to be on the wrong end of the learning curve. The soloprenuer disappears in eighteen months when the maintenance work turns out to be actual work. The client is left with a site nobody knows how to update, a hosting login the freelancer never handed over, a domain registered to a Gmail address that has since been deleted, and a Stripe integration that stopped taking payments six weeks ago because nobody rotated the API key. The soloprenuer moves on to selling AI newsletters. The client is now three months in arrears on the van.

If you insist on doing this anyway

Fine. I am not the fun police and the industry needs new blood. Here is the honest version of the pitch, from someone who has spent seventeen years shipping the thing you are selling. Do this and you'll be at least dangerous rather than actively hazardous.

• →Charge the client for a discovery conversation before you agree a price. If they won't pay for a conversation, they won't pay for the ten conversations that a real project requires.
• →Write everything down. Scope. Deliverables. What's included in maintenance. What isn't. What happens if the site is hacked. Who owns the domain. Who owns the code. Who owns the content. Put it in an email at minimum. A signed SOW ideally.
• →Register the domain in the client's name, not yours. Give them the login. This is not negotiable and I will die on this hill.
• →Get professional indemnity insurance before you take the first invoice. It's £30 a month. Not having it is the single stupidest saving in the industry.
• →Set up a real backup. Test the restore. Do it again next month. A backup you have never restored is Schrödinger's backup.
• →Sell a maintenance retainer or refer the client to someone who does one. A launched site is not a finished site. It is a puppy.
• →Learn what you don't know. Not by reading a Medium article. By breaking a real staging site over and over until the underlying systems stop being magic.

The point

I am not writing this to gatekeep. Gatekeeping is boring and the gate is already off its hinges. I am writing this because when the market corrects, the correction happens to the client, not to the freelancer. The freelancer gets a lesson. The client gets a bill from HMRC they can't pay because the site's contact form was silently dropping enquiries for four months and they only noticed when the accountant did the year-end. The freelancer moves on. The client's business does not.

So if you are one of the thousands of people currently pivoting into web design because you built one site and it went well, please just — sit with the weight of what you're offering. The pretty bit is genuine work and I do not want to diminish it. But the pretty bit is a fraction of the responsibility you are about to accept on behalf of a stranger's livelihood. If you do not know what an SPF record is, you are not ready to be paid to build the digital front door of somebody's only source of income. That is not snobbery. That is arithmetic.

Or, put another way: you would not let a friend who once changed a plug rewire the family home. The website is the wiring. The mortgage is on the other end of it. Act accordingly, or don't act at all.