Conversational Search Is Here. Is Your Business Ready?

Conversational search means asking full questions in plain language and refining the answer with follow-ups, rather than typing keywords and scanning links. The engine replies in sentences, cites a handful of sources and remembers the thread. To stay findable, write pages that answer real questions directly, keep your site readable to crawlers, and check what AI engines already say about you.

Quick links

• What is conversational search?
• How is it different from keyword search?
• Where are people actually doing it?
• What does it change about your content?
• What should you do first?
• Do you need to panic?
• FAQ

What is conversational search?

It is search where you talk to the engine instead of querying it. You ask a complete question, it gives you a written answer, and you carry on from there.

The old pattern was a few keywords, ten blue links and a lot of tabs. The new one looks more like asking a well-read colleague. “Which CRM suits a six-person recruitment firm?” gets an answer. “What about one that works with Outlook?” gets a better one, because the engine remembers what came before.

You will also see it called chat-based search or conversational AI search. Same thing. The label matters less than the behaviour: people have started to expect an answer, not a reading list.

How is it different from keyword search?

In keyword search you match a phrase to a page. In conversational search the engine reads several pages, works out what the person means, and writes one answer from them.

Three differences matter for a business.

Queries are longer and more specific. Google reports that the average AI Mode search is triple the length of a traditional Search query, according to a post by its data science team in May 2026. A person who used to type “accountant Preston” now types the whole situation, including the size of the company and what has gone wrong.

One question becomes several. Google describes a “query fan-out” technique in which AI Overviews and AI Mode issue multiple related searches across subtopics and data sources, so the engine can find a wider and more diverse set of links. Your page is no longer competing for one phrase. It is competing to answer one of the sub-questions behind it.

The answer replaces the click. The engine does the comparing for the buyer. You are either part of the answer or you are not in the conversation at all. There is rarely a page-two consolation prize.

This is why answer engine optimisation exists as its own discipline. The mechanics overlap with SEO, but the goal is different: be the source the engine quotes, not the link a person might choose.

Where are people actually doing it?

Mostly in three places, and the scale is no longer niche.

• ChatGPT. Sam Altman told OpenAI’s Dev Day in October 2025 that more than 800 million people use ChatGPT every week. That is a lot of questions being asked somewhere other than a search box.
• Perplexity. Built as an answer engine from the start, with citations on every response. Buyers doing research tend to like it for exactly that reason.
• Google. AI Overviews sit on top of ordinary results, and AI Mode turns the whole thing into a conversation. Google is not being displaced here. It is changing its own interface.

Each has its own crawler, its own habits and its own quirks. We cover two of them in detail in how to get recommended by ChatGPT and how Perplexity chooses its sources. This post is the map. Those are the street guides.

What does it change about your content?

Your pages need to answer the questions a person would actually ask, in the order they would ask them, in language an engine can lift cleanly.

That sounds obvious. Most B2B sites still do not do it. They are written around what the company wants to say: capabilities, approach, “solutions”. A buyer asking a chatbot about price, timescales or whether you suit a business their size finds nothing to quote.

Write the question, then the answer

Put the question the way a buyer would phrase it in a heading. Answer it in the first sentence or two underneath. Then add the detail. This is the same shape as every H2 on this page, and it is deliberate.

Be specific enough to be quoted

Engines prefer a sentence they can repeat without editing. “Most projects take six to ten weeks” gets used. “Timescales vary depending on requirements” does not. If a number depends on scope, say what it depends on. That is still more useful than silence.

Cover the follow-up questions

Conversations have second and third turns. If your page on a service never mentions cost, the engine will go and find someone who does. Think about what a buyer asks next, and answer that too, on the same page or one linked from it.

Write like a person who knows the answer

Conversational engines reward plain, confident writing. Hedged paragraphs full of “it depends” give them nothing to carry into a reply. If you genuinely know the answer from doing the work, say it, name the exception, and move on. A named author with real experience also gives an engine, and a sceptical reader, a reason to trust the page.

Keep the technical basics intact

An engine cannot quote what it cannot read. Google’s documentation is refreshingly plain on this: a page must be indexed and eligible to be shown in Google Search with a snippet to appear as a supporting link, with no special markup or AI-specific files required. For the other engines, check your robots.txt is not blocking their crawlers and that your main content is in the HTML rather than injected by JavaScript.

What should you do first?

Start by finding out where you stand. Then fix the pages that matter most. You can do the first step this afternoon.

1. Ask the engines. Open ChatGPT, Perplexity and Gemini. Type the questions a buyer would ask, not your brand name. Note whether you appear, whether the facts are right, and who is named instead.
2. Ask a follow-up. Push one step further: add a constraint such as budget, location or company size. The second answer is often where a competitor appears, and it tells you which sub-questions you are missing.
3. List the questions you already get on sales calls. These are your content brief. If the answer is not on your site in plain words, it is not available to an engine either.
4. Rewrite your top service pages answer-first. Question as the heading, direct answer underneath, specifics where you can, a short FAQ at the foot.
5. Check crawler access. Robots.txt, server-side rendering, indexability. Dull, and the one that quietly sinks the rest.
6. Look at your wider footprint. Reviews, directory listings and mentions on other sites all feed what an engine believes about you. Consistency counts for more than volume.

If you want the long version of step one, our AI search audit checklist walks through it in about 30 minutes.

Do you need to panic?

No. Worth a look, though, and sooner than later.

Conversational search has not made classic SEO irrelevant. Google still sends traffic, and the pages that rank well are often the pages engines draw from. Nobody sensible is suggesting you tear up your content plan.

What has changed is the cost of ignoring it. A year ago, not being cited by an AI engine was a missed bonus. For a growing share of buyers it is now the research step, and a business that does not turn up in it is simply not considered.

The encouraging part is that the early moves are cheap. Clearer pages, honest specifics and a site that machines can read also happen to be better for humans. Few of your competitors will have done it yet, which is the useful bit.