The way people search has quietly changed shape. For twenty years the game was keywords: a customer typed “web designer swindon” into a box and skimmed a list of blue links. Now they open ChatGPT, Gemini or Google’s AI answers and ask a proper question, in full sentences, the way they would ask a knowledgeable friend. And the AI replies with a short recommendation, often naming just two or three businesses.
If your business is one of those named, you win the customer before they have visited a single website. If it is not, you are invisible, no matter how good your old keyword rankings were. The businesses that get named are not the ones that guessed the right keyword. They are the ones that answered the actual question the customer asked. This guide shows you how to find those questions, and how to become the answer.
How people search now: whole questions, not keywords
The shift is bigger than it looks. Old keyword searches averaged two or three words. The prompts people type into AI tools run to eight, fifteen, sometimes more, and they carry context, constraints and intent all in one breath.
Compare the two. In 2015 someone looking for software typed “best crm small business”. In 2026 they ask an assistant “what is the best CRM for a small service business with under ten employees that needs email automation”. Same person, same need, completely different search. The second one tells the AI exactly who they are and what they want, and the AI answers with something specific.
This is not a fringe habit. Conversational, natural-language queries now drive more than a third of all search interactions, and Gartner projects that over 80% of online searches will involve an AI assistant by the end of 2026. Since 10 July 2026, Google itself generates an AI answer for effectively every query, so even people who think they are “just Googling” are being handed an AI response first. The full-question search is now the normal search.
Why this is good news for small businesses
Long, specific questions are harder for big generic brands to own and easier for a focused local business to win. “Web design” is a battlefield. “Who can rebuild a slow WordPress site for a Swindon tradesperson without breaking the booking form” is a question a national brand never wrote a page about, and you can answer perfectly.
The catch is that you only get named if you have actually answered that question somewhere the AI can read it. So the work is simple to describe: find the real questions your customers ask, and make sure clear, specific answers to them exist on your site.
Step 1: Map the questions your customers actually ask
You already know most of these questions. They are the ones you answer on the phone, in your inbox and in the first ten minutes of every enquiry. Start there.
- Mine your own inbox and calls. Write down the questions customers ask before they buy: “how much will this cost”, “how long will it take”, “can you work with my existing site”, “do you support it afterwards”. These are search prompts in disguise.
- Ask the AI tools directly. Open ChatGPT and Gemini and ask the questions a customer in your area would ask, for example “who can build an e-commerce website for a small shop near Swindon” or “how do I get my business to show up when people ask AI for a recommendation”. Note what comes back, and whether you are in it.
- Use the “People also ask” and autocomplete trails. Google’s related questions and search suggestions still reveal the phrasing real people use. So do the follow-up questions an AI offers at the end of an answer.
- Add the constraints. Real prompts include qualifiers: budget, location, industry, urgency, platform. “Cheap website” is vague; “affordable website for a startup that I can update myself” is a real prompt. Capture those specifics.
The aim is a plain list of ten to twenty genuine questions, written the way a customer would say them out loud. That list is your content plan.
Step 2: Answer each question fully, in one place
AI assistants quote sources that answer a question completely and self-containedly, rather than pages that only half address it. So for each question on your list, give a clear, direct answer in its own short passage. Put the answer first, then the detail. Use the customer’s own words in the heading, because that is what they are matching against.
For anything sequential or comparative, use a short list or steps. Add a natural FAQ near the foot of your key pages that answers the exact questions people ask out loud. This is the single most AI-friendly shape you can give a page, and it doubles as reassurance for the human reading it. Our guide to writing content that AI cites as a source goes deeper on the exact structure, and FAQ and schema that AI loves to quote shows the markup that helps machines understand each answer.
Step 3: Be specific, because vague answers do not get quoted
The prompts are specific, so your answers must be too. “Websites vary in price” tells the AI nothing it can repeat. “A small business website in the UK typically costs between two and eight thousand pounds depending on the number of pages and features” is a sentence an assistant can lift straight into an answer, with your name attached.
Wherever you can, replace generalities with real figures, timelines, examples and outcomes. Specific, verifiable answers are what make a passage safe for an AI to repeat, and they are what make a wavering customer trust you.
Step 4: Check whether the AI is actually naming you
Answering the questions is most of the job, but you should close the loop. Once a month, ask the main assistants the questions on your list and see whether you appear, and how you are described. If a competitor is named and you are not, look at what they answered that you did not. If your description is thin or wrong, that usually means the AI could not find a clear, current source, so give it one.
This is the same instinct behind checking whether your business appears in AI search and understanding why AI recommends some local businesses and ignores others: you cannot improve what you never look at.
The bottom line
Your customers have stopped typing keywords and started asking questions. The businesses that get named in AI answers are simply the ones that took those real questions seriously, answered them clearly and specifically, and made sure the answers were somewhere a machine could read them. None of that requires a big budget. It requires knowing your customers and writing plainly.
If you would like help finding the questions your customers ask and turning them into pages that get you named, that is exactly the work we do. Take a look at our AI services, or read how to get your website into AI search results for the wider picture. When you are ready, book a free AI visibility chat and we will show you which questions you already win, and which ones are slipping away.
Answer the question your customer actually asks, and you become the answer they get.