Build products, not pages

It's Saturday. You describe an idea to an AI over coffee, and by dinner there's a working app — a form, a database, a way to pay. You never wrote a line of code. Total cost: a domain and an afternoon.

A few years ago that was a fantasy. Now people build an app without coding every weekend, and whole businesses start on a shoestring budget. The expensive part isn't building anymore. It's getting found.

So you go looking for visitors, and the same advice is waiting in every forum: make pages. Thousands of them. One for every city, every "best X for Y", every keyword a tool spits out. Google sends traffic to pages, the advice goes, so more pages must mean more traffic.

Here's the plain answer: no. Ship products, not pages. Testing an idea is cheap now — build many small things, put them in front of people, keep the ones that get used. Mass-producing pages for search is the one move Google names in its spam rules and treats as spam, not as cleverness.

Testing an idea costs a weekend, not a year

Think of a market stall. You set one up on Sunday morning. By evening you know whether people want what you're selling. If they don't, you pack up and try the next idea next weekend.

That's the whole game now. Many stalls. Keep what sells.

Building pages to validate a business idea inverts this. You spend months building a thousand doors that all open into the same empty room, hoping a search engine walks someone through each one. Even when it works, it works slowly — and you've built nothing you'd show a real person with pride.

What Google actually punishes

Picture a printing press in a basement. It runs day and night, producing pages nobody asked for. "Best plumber in Tulsa." "Best plumber in Tacoma." Same three paragraphs, city name swapped. There's no plumber. There's just a page, built because a keyword exists.

Google's spam policies, updated in 2024, have a name for this: scaled content abuse — mass-producing pages mainly to rank in search. It doesn't matter whether a person, a script, or an AI did the producing. And this is the kind of thing that can sink a whole site, not just single pages.

Two neighbors in the same rules:

• Expired domain abuse — buying an old domain for its links and history, then filling it with unrelated or low-value content to rank.
• Link spam — buying links or working link schemes instead of earning them.

If you're tempted by one of those "places to spam your website" lists, that last one is you. Links nobody clicked never brought anyone visitors anyway.

Is AI-generated content bad for SEO, then?

Google has said publicly that it doesn't ban content just because a machine helped write it. The question, in their telling, is whether a page exists for people or only for rankings.

So: one honest page, written with AI help, answering a real question — fine. Ten thousand pages generated overnight around a keyword list — that's the printing press again.

Where exactly the line sits, nobody outside Google can say. Treat "thin, at scale" as the danger zone.

Programmatic SEO: where it tips into spam

A train site with a live timetable for every route generates thousands of pages — and it isn't spam. Every page is real. It shows real times, updates itself, and a person landing from search gets what they came for.

Now the same technique applied to "accountants in [town]" for every town in the country, where the only thing that changes is the town name. Nothing on the page works for anyone. Programmatic just means "built by code from a data set" — whether it's spam depends on what a person gets when they arrive. This version is scaled content abuse with a nicer name.

The test: if search engines vanished tomorrow, would the page still have a reason to exist?

A five-question self-check

Before you build the thing — and before you publish any page around it:

1. Would you build it if Google didn't exist? If the only reason is traffic, it's a page farm with extra steps.
2. What problem does it solve, and for whom? One sentence. If you can't, the idea isn't ready.
3. Do people come back? A tool people reuse is a product. A page read once and forgotten is scenery.
4. What's unique about it? The thing a user would mention when recommending you. If the honest answer is "the keyword," stop.
5. Would you show it to a real user, in person? If you'd be embarrassed, that feeling is data.

Pages that fail questions 1 and 5 are pages for Google, not for people. Make enough of them and it's the spam above.

Where the first visitors actually come from

A brand-new domain is an empty side street at 3 a.m. The best sign in the world, and nobody walks by until the street gets known. People in SEO call the quiet stretch the "Google sandbox." Google has never confirmed a sandbox exists, but new sites waiting months for traffic is common experience.

Either way, waiting for search is the wrong first move when promoting a new product. Your first hundred users come from people, not pages:

• Launch where crowds already gather — see How to launch a product people can find and How to launch on Product Hunt.
• Submit to directories real humans actually read — Startup directories: which are actually worth submitting to sorts the useful from the pointless.
• Show up in communities as a member who builds things, not as a billboard.

If the product survives contact with real people, pages start to make sense: a page per real feature, written to answer a real question. Pages support a product. They don't replace one.

The shortcut you'll be offered: an old domain

The page-farm advice usually arrives with a shortcut attached — buy an expired domain that already has links, and build your thousand pages there.

Be careful twice. A good expired domain is a corner that signposts have pointed to for years: universities, newspapers, Wikipedia. A bad one is a corner where stolen goods were sold — foot traffic, sure, but a past full of spam, casino, pharma. And filling a bought domain with unrelated content to rank is expired domain abuse — the exact move the shortcut suggests.

One honest limit: nobody outside Google can see whether a domain carries a manual penalty. Only Search Console shows that, and only its owner can look.

What to do now

1. Write one sentence: what problem, for whom. Can't do it? Next idea.
2. Build the smallest version this weekend. No-code tools, AI help, whatever makes it real. Rough and used beats perfect and ignored.
3. Put it in front of people this week — launch it, submit it to directories worth the time, or show it to ten people in a community you're part of.
4. Run the five questions over every page you plan to publish, and cut anything that exists only for a keyword. Set up Google Search Console so you can see what search actually sends you.
5. If a domain with history is part of the plan, check its past first: the Wayback Machine, a site: search — or our DD Backlink Strainer Chrome extension, which vets domain lists as you browse auctions (free plan: 250 checks a day).

Ship many. Keep what works. That's the approach that survives every algorithm update — and the one you'd still be proud to explain to a user face to face.