Your website used to be a destination you sent people to. It is now, more often, a source that gets read and summarised on your behalf. That is a genuine change in what the thing is for, and it changes how you should build it.
The numbers behind the shift: zero-click searches, where someone gets their answer without visiting any site, rose from 56% to 69% in one year. Where a generated summary appears, people click a traditional result roughly 8% of the time.
The practical consequence is not "websites are dead". It is that a beautiful site that says nothing specific is now worthless twice over: it does not convince the smaller number of people who arrive, and it gives nothing quotable to the systems deciding whether to mention you at all.
Skip to: what changed · how to build for it · what to stop paying for · what still matters as much as ever
What actually changed
For twenty years the job was: rank, get the click, convert the visitor. Every decision followed from that, including a lot of decisions about looking impressive.
Two of those three steps got weaker. Fewer clicks leave the results page at all, and the pages quoted inside answers are increasingly not the ones ranking highest. One analysis put the overlap between top links and cited sources at under 20%, down from about 70%.
So your site now has two audiences with genuinely different needs. A person, who wants to be convinced. And a machine reading on someone else's behalf, which wants a specific, checkable claim it can repeat without getting it wrong.
The good news is that writing for the second one makes you better at the first. Specific beats vague for both.
Sources: published zero-click search studies 2025 to 2026; Brandlight source-overlap analysis; Gartner projection that roughly 25% of queries run through assistant-style search in 2026.
How to build a site that gets quoted
Answer the question in the first paragraph
Not a welcome message. Not your founding story. The actual answer, in one or two sentences, in plain text, at the top. Everything else can follow it. If a reader has to scroll to find out what you charge or whether you serve their town, so does everything else.
One page per thing you do
The single biggest structural mistake is a combined services page listing nine offerings in nine bullet points. That is one thin page competing for nine different questions. Nine real pages each answer one question properly. This is dull, effective work.
Put the facts in text, not in pictures
Prices in an image are invisible. Hours in a graphic are invisible. Service areas in a map screenshot are invisible. If it matters, it goes in text.
Be specific enough to be quotable
"Quality service you can trust" cannot be quoted by anyone. "Same-day repair across Scottsdale and Tempe, typically $180 to $340, no call-out fee" can. Numbers, place names, timeframes and named constraints are what survive summarising.
Mark up what the page is
Structured data is how a page states plainly what it represents: a business, a service, an article, a set of questions and answers, an event. It is not a ranking trick. It removes ambiguity about the facts you already published.
Make it fast and make it plain HTML
A page whose text only appears after scripts run is a page that can be read wrong or not at all. Static HTML with the words actually in the file is faster for people and unambiguous for everything else. It is also cheaper to host, often free.
What to stop paying for
- A homepage carousel. It has been measured for over a decade and it has never worked. It also buries every specific claim you have.
- Stock photography of people who do not work for you. It is invisible to a reader and worse than nothing to a buyer who can tell.
- A monthly retainer for "content" that says nothing checkable. Twelve articles a year with no numbers, no names and no prices is twelve pages nobody will ever quote.
- Rebuilds that change how it looks and not what it says. If the new site contains the same vague sentences, you have bought a paint job.
A useful test before you approve anything: pick any page and ask, could a stranger repeat one specific, checkable sentence from this page to a friend? If not, the page does not yet do its job, whatever it looks like.
What still matters exactly as much as it did
Plenty. The shift is real but it is not a licence to ignore the basics.
Speed. People still leave slow pages, and slow pages still get read less thoroughly by everything else.
Mobile. Most of your visitors are on a phone, often outdoors, often in a hurry.
Accessibility. A site that a screen reader can navigate is a site whose structure is unambiguous, which helps every reader of every kind. It is also the right thing to do.
Whether it converts. A page can be quoted everywhere and still fail if the phone number is hard to find, the form is long, or nobody answers when someone calls.
Questions people ask
Do I still need a website at all?
Yes, and arguably more than before. It is the one source about you that you fully control, and it is what the sources you do not control are checked against. Without it you are described entirely by other people.
Should I rebuild or fix what I have?
Usually fix. Most sites fail on what they say, not on how they are built. Rewriting the top of each page to answer the question directly, splitting a combined services page, and putting prices into text is cheaper than a rebuild and does more.
Does design still matter if fewer people visit?
It matters more per visitor, because the visitors you do get are further along and more serious. Fewer, warmer visits raise the cost of a confusing page.
How do I know if any of this worked?
Traffic alone will mislead you now, because the wins increasingly happen without a visit. Watch calls and form fills against the questions you want to win, and check whether you are named when those questions are asked.
Want a site that gets quoted?
We build fast, static, specific websites for local businesses: one page per service, the answer at the top, prices in text, marked up so there is no ambiguity about what you do. Fixed price, roughly two weeks.