Content teams are feeling a new version of an old anxiety: how will people find us? For years, it has been Google. Now it’s AI.
The first round of advice to improve AI visibility was technical: schema, markup, knowledge graphs. It all matters, but we’re discovering the most useful work happens in the content itself.
It is somewhat surprising that AI search seems to reward the same disciplines good editors already understand — pages have to make their meaning visible. They have to say what the organisation knows, what the evidence is, how ideas relate to each other, and why that page belongs to this organisation rather than to anyone else.
These are old disciplines — clear headlines and subheads, a lead that plainly states the claim, a ‘so what’ context paragraph that explains why all this might matter to readers.
And AI search appears to reward all of it.
This is already a commercial issue. We are hearing ecommerce teams at major retailers say a double-digit percentage of referral traffic now comes from AI — not Google. And the numbers are rising fast.
When you ask ChatGPT or Claude about something, you can see what they are doing. Answers come back stitched together from pages that made their meaning easy to find with clear headings and specific claims.
So how can you make your corporate content more visible to AI?
Make the page’s meaning visible
The strongest gains don’t seem to come from making hidden technical changes to the page source.
Instead, they come from rewriting the content so relationships are visible to human readers.
That means setting out clearly: what this page is trying to say, what evidence we have, where it fits in this website and which parts of the business offering it refers to.
Tables, summaries and relationship notes help AI tools because they make the structure visible.
But they can’t be hidden in the page source — they need to be clear to human readers too.
Put the source context on the page
On our own website we have started to publish a short block at the foot of the page that names the author, the publication date, the collection the piece belongs to, the service areas it relates to, and where it first appeared.
That block is readable for humans and answers questions a careful reader might ask before deciding how much to trust a page.
But it also gives an AI retrieval system the same clues: who wrote this, where does it sit, what is it related to, and which other pages help explain it.
Decide what each page is meant to prove
That means old-fashioned editorial decisions are important again:
What question is the page answering?
Which service, expertise or audience need does it connect to?
What proof do we have?
Corporate content can sometimes hide its own usefulness.
Articles can discuss the organisation’s support for causes or issues but never say why. Is it because you want potential employees to know? Is it because you want regulatory policy change? Say so.
The same applies to service lines. A thought-leadership piece that names a problem your customers face but never names the service you sell to address it is leaving its most commercially useful relationship implicit. Link the problem to the offering. Say why the offering is relevant.
Pages that surface entity relationships — related topic, related service, related case — perform 30 per cent better in AI retrieval than pages that leave those connections to be inferred.
If the page never makes the business meaning clear, it becomes hard for a person — and for the AI — to know what the story proves.
And that meaning needs to be part of the published work, not hidden in the CMS.