Last spring a guy who runs a bistro a couple of streets from our office bought a website for 4,000 crowns. Someone on a freelance marketplace, three-day turnaround, looked perfectly fine on the demo. Eight months later he was in our inbox. The site crawled on half the phones that opened it, Google had it buried on page four for the one search that mattered - his neighborhood plus the word bistro - and the person who built it had gone quiet. He couldn't even change his own opening hours, because he was never given a login. He paid 4,000 once. Then he paid us 35,000 to start over. That is the whole article in one story. Let me show you where the other 31,000 actually went, because it is the same place it goes for most people who buy cheap.
A cheap website is rarely cheap. The sticker price pays for the one visible thing - a page that looks okay on a demo - and quietly skips the invisible things that decide whether it works: maintainable code, fast load times, a mobile layout that holds together, on-page SEO, security updates, and your own access to the site. Those gaps come back later as real bills: a rebuild, sales lost to a slow or broken site, months of Google invisibility, a security cleanup after a hacked plugin, and the afternoon you discover you don't actually own your own website. Cheap is genuinely fine for a placeholder or a weekend test of an idea. It's an expensive mistake when it's the thing your business runs on. Here is where the money leaks, a three-year comparison of cheap versus done properly, and the newest version of cheap - the AI vibe-coded site - which looks the most convincing of all and fails in the same old ways.
Four things people mean by a cheap website
Not all cheap is the same, and the trap is a little different in each case.
- The marketplace freelancer. A few thousand crowns, fast turnaround, usually one person you found on a listing site. The work can be genuinely okay - or it can vanish the moment something breaks. The risk is not the price, it's that there is no one on the hook afterwards.
- The do-it-yourself builder. Wix, Squarespace, a free WordPress theme. Cheap to start, and fine for a while. The cost shows up as a monthly fee that never ends and a hard ceiling the day you need something the platform doesn't allow.
- The template flip. A cheap agency buys a 40-dollar theme, drops in your logo and text, and ships it as a custom site. It looks like a real website because it partly is one - just the same one as a few hundred other businesses, with SEO and speed left at the factory defaults.
- The AI vibe-coded site. The newest one, and the most convincing. Someone generates a slick-looking site with an AI tool over a weekend. More on why this one is special below.
Where the money actually leaks
The bistro didn't pay 35,000 because we felt like charging it. He paid it because a cheap site leaks money from six or seven places at once, and none of them are on the original invoice.
The rebuild you didn't budget for. This is the big one. When the code is a mess or the platform is a dead end, there is no fixing it in place - you pay again, from scratch. Paying 4,000 and then 35,000 is not 35,000. It's 39,000, plus eight months of a site that was working against you.
The customers who left before they read anything. Cheap sites are slow and they break on phones, where most of your traffic now is. Every extra second of load time drops conversions, and a layout that falls apart on mobile sends people straight back to Google. You never see these losses on a bill - they just don't show up as sales. We wrote about exactly what Google measures here in the piece on Core Web Vitals and why slow pages lose customers.
The SEO you never got. A cheap build skips the boring parts that make Google trust a site: sensible page titles, headings in order, meta descriptions, a sitemap, clean URLs, image alt text. So you sit on page four and wonder why the phone doesn't ring. Fixing it after launch costs more than doing it during the build. Our technical SEO checklist is the list most cheap sites skip entirely.
The security cleanup. The classic is a WordPress site stacked with free plugins that nobody updates. One of them gets exploited, and suddenly your site is redirecting visitors to something you'd rather not describe. Cleaning up a hacked site and hardening it afterwards routinely costs more than the original build.
The site you don't own. On cheap jobs the domain is often registered under the freelancer's name, the hosting is their account, and you were never given admin access. As long as everyone is happy this is invisible. The day the freelancer disappears or you want to move, you find out you were renting your own front door.
Nobody to call. A 4,000 crown price never includes the thing you actually need most: someone responsible when the site goes down at 8pm on a Friday. Ongoing project support is not glamorous, but it is the difference between a two-hour outage and a two-week one.
The three-year math
Put it on a timeline and the bargain evaporates. Rough numbers, Czech market, a small business presentation site over three years:
| Line item | The cheap path | Done properly |
|---|---|---|
| Initial build | 4,000 - 8,000 CZK | 35,000 - 60,000 CZK |
| Rebuild within ~12 months | 35,000+ CZK | 0 |
| Lost sales (slow / broken mobile) | hard to see, easy to feel | minimised |
| SEO fix after launch | 10,000 - 20,000 CZK | included in build |
| Security cleanup (if hit) | 5,000 - 15,000 CZK | avoided by updates |
| Support and small changes | ad hoc, often stuck | predictable monthly |
| Three-year total | usually higher | usually lower |
The cheap column looks great on day one and worse every month after. I'm not saying done-properly is always 60,000 - for a lot of small businesses it lands lower, and we break the real ranges down in how much a website actually costs in 2026. The point is that the gap between the two columns on day one is almost never as big as the gap by year three.
The newest cheap: vibe-coded websites
There is a new entry on the cheap list, and it is the most dangerous because it is the most convincing. AI tools like Lovable, v0, and Bolt can generate a genuinely good-looking site in a weekend, with real code behind it. For the person who built it, and often for the client, it feels like getting a 60,000 crown site for the price of a subscription.
Then it goes to production. In the code we audit from these tools, the pattern is depressingly consistent: security holes, no tests, secrets left in the open, and a structure so tangled that changing one thing breaks three others. It isn't that AI writes bad code on purpose - it's that it writes confident code, and confident-looking broken code is harder to spot than obviously bad code. We laid out what actually turns up in these projects in 10 signs your vibe-coded MVP is technical debt, and fixing them is a whole service on its own - vibe-code rescue. Vibe coding is a great way to start. It's a bad way to finish without someone checking the work.
When cheap is actually the right call
I run an agency, so you'd expect me to say never go cheap. I won't, because sometimes cheap is the smart move.
If you are testing whether an idea has any customers at all, a builder or a vibe-coded prototype is exactly right - spend as little as possible until something works. That is the whole logic behind building an MVP instead of a finished product. A one-page site for a single event, a personal page nobody's income depends on, a placeholder while you save up for the real thing - all fine. The rule is simple: cheap is right when the site is disposable, and wrong when it isn't. The mistake is using a throwaway tool for the thing your revenue runs through.
How to tell if your cheap website is already costing you
Before you spend anything, find out whether your current site is actually the problem. Five minutes, no paid tools.
- Open it on your phone, on mobile data. Time how long until you can read and tap. Over three seconds, or text you have to pinch to read, and you're losing people before they see your offer.
- Google your own business plus your town. If your site isn't the first result, the basic SEO was skipped and even people who know your name can't find you.
- Try to change one line. Log in and edit a phone number or a price. No login, or you have to email someone? You don't control your own site.
- Find out who owns the domain and hosting. If it's not registered to you and paid from your account, you're renting your own front door.
- Ask who you call when it breaks. If the honest answer is nobody, the cheap price left out the one thing you'll need at the worst possible moment.
Fail one of these and it's worth a conversation. Fail three and your cheap website is already the most expensive thing on your marketing budget - you just can't see it on an invoice. That's usually also one of the signs it's time for a redesign.
What to do about it
If your current site passes the check, leave it alone - a working cheap site is a fine cheap site, and rebuilding it would be me making the exact mistake this article warns against. If it fails, don't rush to the cheapest fix again, because that's the loop the bistro was stuck in. Decide what the site is actually for, budget for it once, and make sure whatever you buy is fast, works on phones, has the SEO basics, and is registered in your own name. If you're picking who builds it, the twelve questions here will save you from the second cheap website.
At Kosmoweb we build sites that don't need redoing in a year as part of web development, keep them fast and secure through project support, and clean up cheap or AI-generated sites that already went wrong. If you're not sure which camp your site is in, send us the URL - fifteen minutes, free, and we'll tell you honestly whether it's worth keeping or it's quietly costing you more than a proper one would.