“Getting a website made” is the wrong question
Berger+Team · Aug 27, 2026
“Getting a website made” is the wrong question because it describes a deliverable and not a job: this many pages, a design, a launch date. The question that actually affects your business sounds different: which questions should my business be able to answer around the clock — for people and for machines — and who makes sure those answers are right?
Why the old question no longer holds
The old question comes from a time when a website was above all a place you visited: someone types something in, lands on you, reads, calls. In that world, “getting a website made” really was the right order to place.
Today a growing share of the decision happens before anyone even sees your site — inside a summarised answer an AI assembles from various sources. Your website has picked up a second audience, meant every bit as seriously as the first: machines that read your facts and pass them on. The place has turned into an interface — a digital knowledge and service interface for people and AI agents. How that shift developed over the years is something Florian traced in the post Four phases every website is going through right now on this blog.
That is why plenty of businesses today go looking for a “web agency” and realise mid-conversation that the term no longer covers what is needed. Not because design has stopped mattering. But because what decides success sits underneath the design.
The better question has three parts
First: which questions? Write down the ten to fifteen questions your business genuinely answers every week. When are you open. Do you do this for private customers too. Do you come out to the far end of the valley. How fast can I get an appointment. Roughly what does something like this cost. That list is worth more than any wish-list layout — it is the actual brief.
Second: for whom? Every one of those answers needs two forms: one a human enjoys reading, and one a machine understands unambiguously. That is not duplicate text, it is the same content with a machine-readable structure underneath. Without that structure an AI has to guess — and when it guesses, it sometimes guesses wrong and sounds confident doing it.
Third: who keeps them current? This is the part that shows up in almost no enquiry and matters most in daily practice. Answers go stale: opening hours change, services get added, a number changes. If nobody owns it, the interface goes stale — and with it the picture an AI paints of you.
Why the wrong question gets expensive
The wrong question gets expensive because it shapes a project that ends at launch — exactly where the benefit starts. A deliverable is finished at some point. An interface is never finished; it is either looked after or it is not.
On top of that comes a second cost driver almost nobody budgets for: when the same facts have to be maintained in five places, sooner or later three of them are out of date. What that does in daily practice is something we worked through in the post Every program knows only its piece with the actual numbers.
Five questions to ask instead
Whoever you are talking to, these five questions separate an offer with substance from a merely pretty one:
- Where do my content and facts live, and do they belong to me?
- Are they stored machine-readably, or only styled?
- What can I change myself after launch, without commissioning anyone?
- Are my language versions standalone — or is there just a translator laid over them?
- What happens in the first year after launch, and who is my contact then?
With us the answer to the last question is the shortest one: the same person you spoke to before. A freelance collective has no layer standing between you and the person who actually builds.
What this explicitly does not mean
Not that design is irrelevant — no machine-readable format in the world saves a page that puts people off. Not that you have to rebuild everything — very often it is enough to add structure and maintainability to a foundation that already exists. And certainly not that you need “something with AI”. You need correct answers in one place, in a form both audiences can read.
Where to start in practice
Start with the list. Ten questions your business answers constantly, in one document. Anyone holding that list runs every supplier conversation differently — no longer about page counts and layouts, but about what the business should be able to do.
If you would like to go through that list together before you ask anyone for a project budget: get in touch. That is exactly what the first conversation is for.
Frequently asked questions.
How do I tell whether an agency genuinely understands how AI reads my website?
A simple test: ask about their last three projects and how AI visibility was handled. If the answer stops at meta tags, page speed and classic rankings, that only covers half the job — important, but not decisive for AI answer systems.
Anyone who genuinely understands how AI reads a website talks about more concrete things: a machine-readable profile such as `llms.txt`, structured data in JSON-LD format for facts like opening hours or services, and whether AI crawlers and agents actually have technical access to the site. These are the levers that decide whether an AI can cite or recommend your business correctly.
What sets you apart from an agency?
At a classic agency, your request lands in a ticket system first, then with a project manager, who passes it to a developer you never actually talk to. With us, you email or call the person who actually writes the code — usually Florian himself.
Concretely: a change request doesn't go through three hand-offs, it goes straight into the work — often discussed and scheduled the same day. No management layer, no middleman: what you pay goes into the result, not into coordination.
How does a project run — from first conversation to go-live?
It starts with a no-obligation first conversation: what do you want to achieve, what already exists, what is missing? From that comes a clear proposal with scope, process and a realistic estimate of the timeline. The concept phase follows — content, page structure, design direction — before implementation begins.
During implementation you see interim states instead of a black box, and your feedback flows straight into the work. Before go-live we check content, languages, technology and redirects together. After launch you get an introduction to maintaining your content — and, if you wish, ongoing support to keep the website current and visible.
Do I need to rebuild my website for AI visibility?
Not necessarily. Structured data, clean multilingualism and performance can be retrofitted. But if the foundation is an old tower of plugins, a foundation that brings these things built in is often the faster and cheaper way.
