Content teams are feeling a new version of a very old anxiety: how will people find us?
For years, it was Google. Now it’s AI.
Until recently, thinking about AI visibility went a lot like SEO: schema, markup, knowledge graphs.
Now, it’s starting to look like most important stuff happens in the content itself.
It’s fast becoming mission-critical — we are hearing major retail e-commerce teams report that a double-digit percentage of traffic is now coming from AI instead of search.
So how do you make your content available and useful to AI?
Research says there’s a few ways that stand out:
Tell AI what your page is about
For all their genius, the AIs are still a bit dumb at figuring out what content is for and how it all fits together.
That means you need to spell it out.
Surprisingly, the best gains seem to be found by spelling it out in a way humans can also understand.
That means you can’t bury information about the page in the metadata like we used to with social media but instead need clear signposts on a page that tell the reader (and the AI) what this page is for, where it fits in the scheme of things and what parts of the business it relates to.
It sounds a bit silly and sometimes it can be but it means literally having a box on each page that says who wrote it, when, what collection it belongs to, the service areas it relates to, what other pages it is related to and so on.
The idea comes from a piece of research that shows structured data surfacing entity relationships (topic, service, case, business line etc) improves AI answer accuracy by 30 per cent — but only if it’s also visible to a human reader.
It’s not fully clear why that’s true — but it seems to be something to do with the way AI fetches body content and converts it to plain text, potentially missing or overlooking some metadata.
Use the old-fashioned newspaper tricks
Newspapers have a design concept of ensuring multiple ‘points of entry’ on a page.
The idea is that because a block of text is hard to read, you should break it up with headlines, subheads, images, breakout quotes and graphics.
It’s the reason you can scan a well-designed print newspaper quickly and get a sense of what’s news. Each element works together to tell a little part of the story. If you want more after reading the headline, you can scan the subheads, or the caption on the image, or some other point of entry.
Weirdly, AI needs that too.
Breaking a page up with multiple points of entry gives the AI useful little blocks of information to work with and help it work out what question you are answering.
It also helps the AI cite you properly and surface the right information to answer the user’s question.
For digital content, it means all the old-fashioned editorial disciplines are back in vogue — clear headlines, useful subheads, descriptive captions, FAQs, summaries and ‘so what’ paragraphs, lists and tables.
Include evidence and quote your experts
When AI conducts research to answer user questions it has to make judgments similar to how a human would act: does this information answer the question and is it credible?
Content that works for AI looks an awful lot like traditional journalism — a lead that answers the reader’s question, evidence from multiple sources, and a clear ‘so what?’ statement of context.
Sources can be your own documents, interviews with your subject matter experts or external research.
But they should be explicit in the copy — not footnoted at the end.
Direct quotes from experts work well here, telling the AI that this is not just your view but also the view of a named authority on the topic.
In fact, research shows using direct quotes in your copy is the single best way to boost your content’s ranking in AI answers — lifting AI visibility by more than 40 per cent.
Numbers are also useful — replacing qualitative statements with actual statistics lifts AI visibility by 30 per cent.
Think ‘answer chunks’ not just pages
It’s tempting to have separate, linked pages for topics so they can be updated easily.
But when organising content for AI retrieval it’s better to think in terms of ‘chunks’ or ‘blocks’ of answers grouped on a page that explain separate but related ideas.
A chunk might have a descriptive heading or question, an answer immediately below it in a full sentence, any evidence or supporting material and whatever context is needed to make the information understandable.
It should be grouped on the same page as other content chunks that answer similar questions.
Remember that AI isn’t carefully reading your whole page and intending to faithfully reproduce it — it’s hunting for the quickest plausible answer it can extract.
As a rule of thumb, if a human reader would expect the content chunks to be found together, keep them together.
FAQs are a good example of organising content like this — but it also holds true for product and service pages, support and insights.
Arranging your content in chunks that make sense and aren’t just blocks of body text makes it easier for AI — and people — to read.
Use YouTube for video
It feels a bit unfair, but Google-owned YouTube is consistently one of the best cited sources in Google search’s AI Overviews.
Preferencing your own channels is nothing new, but Google’s highlighting of YouTube videos may be less about unfair competition and more to do with the fact it has backend access to content on a platform it owns rather than having to index videos hosted on a third-party platform.
That means transcripts, traffic patterns, engagement signals and watch patterns are available to Google’s AI — and it’s possible that it simply feels more confident that it’s giving a good answer when it uses YouTube as a source.
All that means there’s a fair argument for simply shifting your video effort to YouTube and embedding its streams in your own channels.
But video cannot stand alone.
Despite the above, the AI’s aren’t actually watching all the millions of videos posted each day — they rely instead to a large extent on the written material around them.
That means videos need words around them — good titles, clean transcripts and clear explanations of what they are about.
Articles based on the material in the video are a good way to do this — quoting what your subject matter experts are saying.