Being quoted by AI answer engines: what actually makes a page citable

This is the part of search that has changed most in twenty years, and it is worth being precise about what it is and is not. An answer engine does not rank ten links and let the reader choose. It reads a handful of pages, writes a single answer, and names some of what it used. If your page is not among the handful, you are not on the page at all.

The work of being one of those sources has picked up a few names, answer engine optimisation and generative engine optimisation among them. The names matter less than the underlying question, which is simple: what makes a page easy for one of these systems to find, understand, trust and quote?

In broad terms, most of these systems do some combination of two things. They search, using a conventional index, and they read the pages that come back. Some also draw on what the model already absorbed during training, which is older and cannot be influenced directly.

The practical consequence is that ordinary search visibility still matters a great deal. A page that no search index surfaces for a question is unlikely to be retrieved and read when someone asks an assistant that same question. Answer engine work sits on top of solid technical search foundations, not instead of them.

The second consequence is that the page has to be readable by a machine on its own terms. A page whose content only appears after a script runs, or sits behind a consent wall, or is an image of text, is a page that may be retrieved and then quietly skipped in favour of one that was easier to use.

The single most useful change most businesses can make is structural. State the answer to the question directly, in a short self contained passage, near the top of the page, and then explain, qualify and expand underneath.

This helps because a system assembling an answer is looking for a passage it can lift and attribute without having to reconstruct your argument from six paragraphs scattered across the page. It also happens to help human readers, who behave much the same way.

Use the question as a heading, in the words a person would actually say. Then answer it immediately in a sentence or two. Then go into the detail. A page built as a series of clearly headed questions and direct answers is far easier to draw from than the same information written as continuous prose.

Vague marketing copy is almost useless as a source. There is nothing in it to quote. Specifics are what get picked up: the actual steps, the actual numbers, the actual conditions under which something applies, the actual exceptions.

That means saying things like which areas you cover, what a service typically includes, how long something usually takes, and what it does not cover. Concrete, checkable statements are the raw material an answer is built from.

It also means being clear about who is saying it and when. A named author, a visible published date and a genuine update date all help a reader and a machine decide whether to rely on the page. Pages with no author and no date are easy to pass over in favour of ones that have both.

Some of this is basic and often missing.

Answer engines, like search engines, work partly on entities: the idea that your business is a particular thing in the world, with a name, a location, a field of work and a set of connections. The clearer and more consistent that picture, the more confidently a system can say your name.

In practice that means your business name, address, phone number, description and areas of work should match across your own site, your Google Business Profile, Companies House, your social accounts, trade bodies and any directories you are in. It also means having a proper about page, real named people, and clear statements of what you do rather than an aesthetic homepage with three words on it.

What is said about you elsewhere matters too, and you influence it less directly. Being described accurately on other people's sites, in trade press, in supplier listings and in reviews adds to the picture. This is slow work and it is not a trick.

Some sites are invisible to answer engines because somebody, often a previous developer or a security plugin, blocked the relevant crawlers in robots.txt or at the firewall. It is worth actually looking rather than assuming.

This is a genuine decision, not an obvious yes. Some businesses do not want their content used this way. Our view is that if you are trying to be found by buyers who now ask assistants, blocking those assistants is working against yourself, but it is your call and you should at least make it deliberately.

There is also an emerging convention of publishing a plain text file describing your site for these tools. It is not a standard anybody is obliged to honour and we would not oversell it. It costs very little to publish one, and we do on our own site.

Hidden instructions aimed at models, keyword stuffing dressed up as an answer, thin pages generated in bulk, and fake statistics all show up sooner or later, and the cost of being caught is higher than the gain. We would rather build something that deserves to be quoted.

There are also no guarantees available here. Nobody can promise you a mention in an AI answer, the systems change frequently, and anyone who tells you otherwise is selling something. What can be done is to make your pages genuinely easy to find, read, understand and trust, which is the same thing that has always worked and is now more clearly rewarded.