When someone asks ChatGPT for the best holiday accommodation in Albany, it does not return ten blue links. It gives one answer, names two or three businesses, and moves on. If you are not in that answer, you do not exist for that person.
This is answer engine optimisation, and it is now part of every site we build. It is not a replacement for SEO. The answer engines lean heavily on what Google and Bing already rank, so if you have no search visibility you will have no AI visibility either. But ranking is not enough on its own. A page can sit at number one and still never be quoted, because it is written in a way that is hard to lift a clean answer from.
What an answer engine is looking for
Strip away the mystique and the model is doing something simple. It has a question, it pulls a handful of pages it trusts, and it looks for a passage that answers the question directly. Then it names the source. So the things that matter are the things that make that job easy.
Lead with the answer. If the page is about what a virtual power plant is, the first sentence under the heading should say what a virtual power plant is. Not a story, not a brand statement. The answer. You can earn the rest of the page afterwards.
Be specific and concrete. Models quote numbers, places, names and dates. “Three bedrooms, 200 metres from Middleton Beach, sleeps six” is quotable. “A beautiful home away from home” is not.
Use the name people use. If your company, product or framework is described five different ways across the web, the model will hedge or get it wrong. Pick one name and one description and use it everywhere: your site, your LinkedIn, your directory listings, your press.
Mark it up. Structured data is how you tell machines what a page is about without ambiguity. Organisation, Product, FAQ, LodgingBusiness, Article. It costs almost nothing and removes the guesswork.
What we did for BnB Collection
BnB Collection has a dozen properties across Albany and Denmark. Each property page now opens with the plain facts an assistant needs to recommend it: what it is, how many it sleeps, how far from the beach, what is across the road. Underneath, lodging schema carries the same facts in a form a machine can read. The result is that guests now turn up saying ChatGPT recommended a specific townhouse by name.
How to check whether it is working
Ask. Open ChatGPT, Perplexity and Google with AI Overviews on, and ask the questions your customers ask. Do it monthly and keep a record. Rankings tell you whether you are in the pool. Citations tell you whether you were chosen. We track both for every client, because the second number is the one that turns into bookings.
See this in practice: Driving Green, BnB Collection and Hockey Stick Advisory.