Let us assume you have done the technical work. Your pages render properly, crawlers can read them, you have a sitemap, your titles are unique, your structured data is in place, you are on your own domain.
And you still have almost no visitors.
This is an extremely common place to end up, and it is the point at which many people conclude that SEO does not work, or that they need to spend money on ads, or that they should give up on distribution and go and build something else.
The actual explanation is usually much simpler, and it is this: your app is one page, and a page can only be found for the things it talks about.
The arithmetic of one page
Sit with the mathematics for a moment, because it is stark.
One page means one topic. One topic means, realistically, one search query you could plausibly appear for. And that one query is essentially always the most competitive phrasing in your entire category, something like "invoice app" or "AI hairstyle tool" or "habit tracker", where you are competing with businesses that have existed for a decade, have thousands of links, and employ people whose full-time job is to hold that position.
You will not win that query. Not this year, not with a domain registered last month, and not by improving your title tag.
But here is the part that should cheer you up considerably. That query is not where the people are. It is where the aggregate is, and aggregates are misleading. The people are spread across hundreds of much more specific questions, each of which is far easier to win, and most of which nobody is competing for at all.
People search for their problem, not for your solution
This is the single most important idea in this article, and it is the one most product builders get backwards.
Somebody with a problem does not know what the solution is called. That is precisely why they have the problem. They cannot search for a category they have never heard of.
So they describe their situation, in their own words, and hope something useful appears.
- Somebody unhappy with their hair does not search "AI hairstyle tool". They search "what haircut suits a heart shaped face".
- Somebody drowning in receipts does not search "expense management software". They search "how do i keep track of receipts for taxes self employed".
- Somebody with a messy spreadsheet does not search "data cleaning tool". They search "how to remove duplicates from two lists in excel".
Your homepage answers the question "what is this product". Nobody asks that question. It is not a question, it is a description, and descriptions do not get searched for by people who have never heard of the thing being described.
Every page you have that answers a real question is a door. Right now you have a wall with your logo painted on it.
What this looks like when someone does it properly
Kurli is a small AI hairstyle advisor. You upload a selfie, your face is mapped in the browser to work out its shape and proportions, and the cuts that suit it are rendered onto your own photo.
The domain, kurli.me, was registered on 27 July 2026, which is a matter of public record in the domain registry.
Now, the obvious way to build that product is one page. A headline, a demo, an upload button, some feature bullets. Most people would have stopped there, and it would have looked completely fine.
Instead the site has guides organised by face shape. By hair type. By length. Separately for men and for women. Sections on hair colour, on bangs and fringes, on thinning hair, on styles that work with glasses, on beards.
Every one of those corresponds to something a real person genuinely asks.
In the first 25 days of that domain's life, ChatGPT sent it 56 visitors and Google Search sent 23. On a domain less than one month old, the AI assistant was the larger source by more than two to one. The absolute numbers are modest and we would not dress them up, but the ratio on a brand-new domain is the striking part. The full breakdown is in this case study.
It happened for an unglamorous reason. When somebody asked an assistant "what haircut suits a heart shaped face", there was a page that answered exactly that. Not a homepage claiming to be a hairstyle tool. A page answering the question.
Why this matters more with AI assistants than it ever did with Google
There is a reason this approach is more powerful now than the same advice would have been three years ago.
Google ranks largely on accumulated authority. Links, history, whether people who found you were satisfied. A new domain has none of these, which is why the standard advice is to wait several months, and why that advice was reasonable.
An AI assistant is doing something different. It has a question in front of it and it is looking for the clearest answer. It does not have the same instinct to distrust a page because the domain is young. If your page is genuinely the best explanation of the thing being asked about, its age is not the deciding factor.
So the strategy of building a specific page for a specific question, which used to pay off over a year, can now pay off in weeks. That is a real change and it is worth acting on while most people are still building one page and waiting.
How to find your questions, for free, in an afternoon
You do not need a keyword tool for this. You need a couple of hours and a willingness to think from the position of somebody who does not know what your product is.
Step one: write twenty questions from your own head. Questions about the problem, not about the product. Force yourself to use the words of somebody who does not yet know the vocabulary of the solution. If you keep writing questions that contain your product category, you are still thinking from the wrong side.
Step two: expand each one with Google itself. Type your question in and look at the "People also ask" box. Those are real adjacent questions. Then scroll to the bottom and read "Related searches". Both are free lists of what people genuinely type, and between them your twenty becomes forty.
Step three: ask the assistants. Take your five most important questions and put each into ChatGPT. Read the answer. Note which sites get cited. That is your real competition for the channel that is currently easiest to enter, and it is often not who you expected. Frequently it is a forum thread or a small blog, which should encourage you.
Step four: pick the ten you can genuinely answer better than whoever is being cited today. Not ten you can produce something about. Ten where you honestly believe you can write the best available answer.
What a real page looks like
This is where people go wrong in the other direction, so it needs saying clearly.
A real page is not three hundred words of filler with a keyword in the title. Producing thin pages at scale is a well-documented way to get your entire site treated as low quality, and the damage lands on the whole domain rather than on the weak pages alone.
A real page for "what haircut suits a heart shaped face" explains what a heart shaped face actually is, how to tell whether you have one, which cuts flatter it and the reason why, which to avoid and why, what to ask your stylist for, and what changes if your hair is fine, or curly, or growing out. It reads as though a person who knows the subject wrote it, because one did.
Then, near the point where the reader is thinking "all right, but which of these is right for me", it links to your tool. Somebody who has just read four hundred words about their own face shape is in an excellent frame of mind to upload a photo.
That is the whole mechanism. Answer the question properly, then offer the thing that goes further.
What to expect, honestly
Ten good pages will not produce a flood. This is not that kind of work, and anybody telling you otherwise is selling something.
What ten good pages produce is a site that answers ten real questions instead of zero, which is a categorically different object from the one you have now. Some of those pages will do nothing. One or two will quietly become the reason people find you. You will not reliably predict in advance which is which, which is an argument for writing ten rather than agonising over one.
Google will take weeks to months. The assistants may be considerably faster, as they were for Kurli. Neither is instant.
And unlike posting on social media, which stops working the moment you stop posting, these pages keep working while you sleep, and they compound, because a site with thirty useful pages is treated differently from a site with one.
The short version
Fixing your rendering, your sitemap and your metadata makes you readable. It does not make you findable. Readable is a prerequisite, not an outcome.
Findable comes from having pages that answer the questions people actually ask about the problem your product solves, in the words they actually use, before they know what the solution is called.
Most vibe-coded apps are one page describing a product. The ones that get found are a set of pages answering questions. That is very nearly the entire difference, and it is available to anyone willing to spend a few afternoons writing.
Not sure which questions are worth writing?
This is the part where an outside view helps most, because it is genuinely hard to think about your own product from the position of somebody who has never heard of it. We are happy to go through it with you and work out which ten questions are worth your time.
Related reading: