The customers who never knew to search for you
Traditional search has one way in. Somebody knows roughly what they want, types it into Google, and you either rank for that phrase or you do not.
That is still the most valuable entry point there is. Nobody is a better prospect than a person searching for exactly what you sell.
But it has always left a whole category of business stuck. What if what you offer is genuinely useful, and people simply do not know it exists? You cannot rank for a search nobody is making. If nobody knows the name of the thing, nobody types the name of the thing.
Historically my advice to those businesses was to forget search and go and buy attention somewhere else. Social ads, mostly. Interrupt people, because you cannot catch them.
That has changed, and it is the most interesting shift in search I have seen in years.
People do not have conversations with Google
Here is the mechanical difference, and everything else follows from it.
In traditional search, people type keywords. Two or three words, sometimes four. And crucially, you can only type a keyword for something you already know exists. If you do not know what the thing is called, or you do not know a solution exists at all, you do not search badly. You just do not search.
Now people are having long conversations with ChatGPT. They are typing whole paragraphs into Google. They are describing a problem rather than naming a product.
Which means for the first time, search has become a place where people discover things rather than just retrieve them. Somebody explains their situation, and the AI says “have you considered doing it this way,” and they learn a solution exists that they never knew to look for.
Traditional search found you what you were already looking for. AI search regularly tells you about something you did not know existed.
Why this matters for every business, not just obscure ones
It is tempting to file this under “useful if you sell something niche.” It is broader than that.
Every business has customers who arrive by a route they never articulated. The person who searched “why is my heating bill so high” and ended up needing a new boiler. The person who asked “should I be worried about damp in a 1930s house” and ended up needing a survey. The person asking “how do small businesses handle enquiries out of hours” who has never heard of the thing that would solve it.
Under keyword search, none of those people found you, because they were not searching for what you sell. They were describing a problem.
Under AI search, that conversation can end with your name in it. But only if there is something on the web, written down plainly, that connects that problem to your kind of solution.
That connection is the opportunity, and it is almost entirely unclaimed.
How to find your entry points
The practical bit. Two steps.
Step one: write down the conversations, not the keywords
You need a list of the things people might actually say to an AI before they end up needing you. Not keywords. Sentences.
There is no reliable data on what people type into ChatGPT, and I am suspicious of anyone claiming otherwise. Keyword tools help a bit, forums help a bit. But the best source is you.
You already know what people say when they ring up. You know the confused first question, the wrong terminology, the thing they thought the problem was before you told them what it actually was. That is the raw material, and no tool has it.
Write thirty of them. Phrase them how a real person would, not how a marketer would.
- Not “emergency plumber Blackpool”
- But “water coming through my kitchen ceiling on a Sunday, who do I even call”
Step two: find out what AI reads to answer them
Then go and look at what actually happens when someone asks those questions.
Run them. See which pages get cited before the AI reaches its recommendation. There are tools that track this properly, and the space is moving fast enough that I would rather you picked a current one than took a recommendation from a blog post.
What you are looking for is the source material. Which specific pages is the AI reading before it decides who to put forward? That is not a mystery you have to guess at. It is observable, and most businesses have never looked.
What to do with what you find
Here is where I differ from some of the advice going round.
You will see people suggest chasing the pages that get cited. Emailing them for a mention, leaving comments, getting yourself onto the listicle. Some of that is ordinary PR and perfectly fine. Quite a lot of it slides into manufacturing mentions, which Google names directly in its guidance and which has a habit of ending badly. I have written about why I leave that alone.
The better use of the information, and where the overwhelming majority of the opportunity sits anyway, is simpler.
Look at what is being cited, and ask whether you could write something genuinely better on your own site.
Usually the answer is yes, and often it is embarrassingly yes, because the cited page is a thin listicle assembled by somebody who has never done the job. You have twenty years of doing the job.
And this is the interesting part
What you will find is a list of topics you would never have written about under the old rules.
Very specific, slightly odd, narrow subjects. Questions no human was typing into Google, so no keyword tool ever suggested them, so nobody in your trade has bothered. But questions the AI generates and searches for on its own, after a conversation with a real person.
That is the gap. Not the big obvious keywords everyone is fighting over. The strange, specific stuff sitting underneath the conversation.
Why this shapes how I run INKON
I will be straight about the connection, because it is the reason I built the business this way.
The old agency model was a one-off build. You paid, you got a site, the content was decided at the start and then frozen. That worked when the entry points were a short, stable list of keywords you could identify once.
Or, if you were given the keys, something else happened. You went off and added your own content, quite reasonably, because you know your business better than any agency does. And somewhere in the process the careful bit got undone. A heading rewritten, a page title changed to something snappier, a location dropped from a paragraph because it read a bit repetitively. All perfectly sensible edits. All quietly dismantling the thing that was making the page findable in the first place.
That is nobody’s fault. It happens because the reasoning was never written down. You were handed a site, not an explanation.
Neither version works now. The entry points are conversational, they multiply, and they shift as people’s habits change. Finding them is an ongoing job and writing for them is an ongoing job. A site built once and left alone cannot chase a moving target, and a site being edited without the reasoning behind it will drift away from one.
So the monthly model is not a payment plan dressed up as a subscription. It is what the work actually requires. Content going out regularly, aimed at the conversations people are having this month rather than the keywords that mattered two years ago.
And there is no gap between what I write here and what gets published for clients. Everything on this blog, the answer-first structure, the plain sentences, naming the town and the service, covering the whole cluster of questions rather than the one obvious keyword, is the same thinking that shapes the content going onto client sites. I am not keeping a better method back for paying customers. The advice is the method. What clients are buying is somebody doing it every month, consistently, without it landing on their to-do list.
Every business I look at has entry points it has never claimed, usually because nobody thought to write about something so specific. That is the whole opportunity, and it is available right now precisely because it looks like too much bother.
Want to know which conversations you are missing? Get your free audit, or email hello@inkon.co.uk. I read every message myself.