← All insights
AI and emerging technology Content and search

Your website now has two audiences

People still visit websites. AI systems increasingly interpret them first. The answer is not a bag of GEO tricks; it is clearer thinking, stronger evidence and content worth citing.

Written by David Carter
3 min read Share on LinkedIn ↗
Designer reviewing a website interface on a laptop

Your website has a new reader. It never gets tired, rarely admires the animation and has no patience for a paragraph that takes 200 words to locate its point.

AI search systems and browsing agents increasingly interpret web content before a person reaches it. They compare claims, extract facts and assemble answers from multiple sources. That has prompted a small gold rush around “GEO”, “AEO”, special files and assorted tricks designed to charm the machines.

Here is the awkward bit: the tricks are not the strategy.

Google’s current guidance says the fundamentals still matter. Its generative search features draw on indexed web content, use retrieval and may run several related searches to build an answer. Google explicitly advises brands to create unique, non-commodity content and says it does not need an llms.txt file, microscopic “content chunks” or copy rewritten specially for AI.

In other words, there is no magic robot seasoning.

There is, however, a communications job to do.

Make the proposition extractable

First, make the proposition extractable. A visitor should be able to identify what you do, for whom and why it matters without decoding six abstract nouns. If the homepage says you “enable transformative outcomes across connected ecosystems”, neither a buyer nor an agent has learned very much.

Second, publish things that only your business can credibly say. Original data, experienced opinions, customer evidence, useful comparisons and named expertise give a search system a reason to cite you. A generic article assembled from the same ten search results gives it a reason to skip you.

Third, build a visible evidence chain. Connect the claim to the method, the method to the result and the result to a person or source. Case studies should contain decisions and outcomes, not just a happy client adjective. Insight articles should show judgement, not merely summarise a trend.

Fourth, design for comprehension. Clear headings, descriptive links, useful image text, logical navigation and fast pages help people, search engines and assistive technology. This is not a technical clean-up after the creative work. It is the creative work.

Finally, stop treating discovery as a traffic contest. A buyer may encounter your thinking inside an AI answer, a social post, a journalist’s article or a colleague’s summary before visiting your site. Consistent language and evidence across those surfaces matter more than forcing every journey through a keyword-shaped door.

Commission what competitors cannot copy

This changes how content should be commissioned. Begin with the decision the reader is trying to make, then identify the expertise only your organisation can contribute. Interview the person closest to the work. Capture the uncomfortable detail: the option you rejected, the assumption that failed, the trade-off the customer accepted. That is the material competitors cannot reproduce with a prompt and a tidy template.

Then test the page like a sceptic. Can a reader find the answer quickly? Can they distinguish evidence from opinion? Is the author identifiable? Do the links prove the claim or merely decorate it? If the page is technically optimised but intellectually empty, the machine may parse it perfectly and decide it has nothing worth repeating.

The web is becoming less predictable, not less important. Your content now needs to be easy to find, easy to interpret and good enough to repeat.

Write for humans. Structure for machines. Give both something worth taking away.

Make the evidence visible

Show the work, not just the claim.

Interfaces, journeys and results give people and machines something concrete to understand and repeat.

Selected Clockwork client work.

Sources and further reading

  1. Google guide to optimising for generative AI features ↗
  2. Google guidance on AI features and websites ↗

Where this becomes useful

Be the source, not the summary

See how Clockwork builds technology stories

Keep thinking