How the AI chooses who to mention
When a customer asks a question, the AI pulls three things together into one answer. What you write yourselves. What others write about you. And what you're known for. Ahrefs studied 75,000 brands. What correlated most closely with being mentioned in Google's AI answers was how often the brand was mentioned elsewhere on the web.
So an AI answer is a recommendation built on your reputation. That's what brand work has always been about. Now there's just a machine reading through all of it every time someone asks.
Four levers make the biggest difference.
1. A clear answer to what you're known for
The AI can only recommend you for something it associates you with. If your homepage says "solutions for modern businesses", it has nothing to tie you to.
Choose two or three things you want to be the obvious answer to. Who do you help, with what, and where? Write it in a sentence a customer would use themselves. A machine can understand and repeat "Accounting software for Danish tradespeople". It can't do anything with "Digital transformation".
Test it yourselves. Ask ChatGPT, Gemini and Google's AI: "What is [your name] known for?" The answer shows what the web as a whole says about you. If it doesn't match what you'd say yourselves, this is where you start.
2. Content that answers customers' questions
People don't search in ChatGPT. They have a conversation. The questions are long and specific, and Google's AI breaks them down into a series of sub-questions before it writes a combined answer. If you have the best answer to one of them, you can get included.
That takes far more content than ten pages about what you sell. Here's how to build it: content marketing. The questions are already in-house:
- Ask sales and customer service what they get asked every week. Keyword research shows the rest.
- Read through quote requests and e-mails from new customers.
- Write down the questions that come up just before a purchase: price, delivery time, the difference between you and the alternative, and when you're not the right choice.
Write one clear answer per question, and give the answer in the first sentence. Use your own experience, figures and examples. That's what the AI can't find anywhere else. Honesty counts too. An answer about who you're not a good fit for makes the rest of your content more credible.
Keep it fresh. If there's a price from 2023 on your site, that may be the one the AI repeats to your next customer. Go through the most important answers every quarter.
3. Coverage where your customers read
What others write about you carries more weight than what you write about yourselves. It always has with people. Now it applies to the machine too.
Find out where your customers get their knowledge. It might be trade media, review sites, podcasts, LinkedIn, professional associations or your partners. Make a point of being mentioned there. Ask happy customers for a review, and reply to the ones you get. Share your knowledge in the media your customers follow. Let your own people talk and write about what they're good at, with their name and title.
This is PR, and it can't be tracked click by click. Even so, this is where AI and branding meet. When anyone can produce content with AI, the winners are the ones people know and trust.
4. The same story on every channel
The AI compares what it finds. If the website says one thing, the Google profile another and LinkedIn a third, the picture of you gets blurry. It's hard to recommend a brand you can't describe.
Look at your channels through the customer's eyes. Are your name, address, what you do and who you do it for the same on the website, the Google profile, LinkedIn and the review sites? Are prices, opening hours and product data correct everywhere? Do your managers and employees tell the same story when they write about your company? It sounds obvious, and it's one of the quickest levers you have.
Four pitfalls
- Tricks for the robots. Special AI files, code hacks and copy written for machines. Google itself says in its guidance that there are no extra requirements for appearing in AI answers.
- Chasing a ranking in ChatGPT. The answers change from one time to the next. In a study by SparkToro, the chance of getting the same list of brands twice was less than 1 in 100.
- Content that sounds like everyone else's. Copy written by AI without your experience gives the AI nothing new to cite.
- Forgetting the foundations. A machine needs to be able to read your pages, and the text needs to be actual text. It's quick to put in place and needs to be right before the rest works.
How to measure whether it's working
Brand takes time, so measure over quarters. Make a list of 20 to 30 of the questions your customers ask, and put them to ChatGPT, Gemini and Google's AI every quarter. Count how often you appear. A single answer tells you nothing. The pattern across many answers does.
Also track searches for your name and the number of enquiries. Ask new customers where they heard about you, and give them "ChatGPT or another AI" as an option. And set up AI traffic as its own source in Google Analytics, so it doesn't disappear into direct traffic.
How we work with it at Amplify
We start with the brand. Before we write a line, we agree with you on what you should be known for. And who you should be the obvious answer for. That happens in the strategy sprint and ends in a plan with clear goals.
Then we build the content plan on the questions your customers ask. Branding, content, social media and coverage pull in the same direction, so the story is the same wherever a customer or a machine comes across you. SEO is part of the foundations. Our own AI agent keeps an eye on accounts and numbers and flags when something needs looking at. The specialists decide what happens.
No one can promise you a place in AI answers. The machine chooses for itself, and the answers change. What you can control is whether there's a clear and credible picture of you when it looks.
Frequently asked questions about being mentioned in AI answers
How does an AI choose which brands to mention?
It pulls together three things: what you write yourselves, what others write about you, and what you're known for. A study of 75,000 brands found that what correlated most closely with being mentioned in Google's AI answers was how often the brand was mentioned elsewhere on the web.
Can you optimise your way to a spot in ChatGPT?
No. The answers change from one time to the next. In a study by SparkToro, the chance of getting the same list of brands twice was less than 1 in 100. Measure how often you're mentioned across many questions, rather than a ranking.
How can you tell whether customers are coming from AI?
Set up traffic from AI assistants as its own source in Google Analytics. Otherwise a lot of it shows up as direct traffic. Also ask new customers where they heard about you, and give them "ChatGPT or another AI" as an option.
Does it require a special technical setup?
No. Google itself says in its guidance that there are no extra requirements for appearing in AI answers. The pages need to be readable, and the text needs to be actual text. The rest is content and coverage.
Your next step
Ask ChatGPT, Gemini and Google's AI three questions:
- What is [your name] known for?
- Who is best at [what you sell] in Denmark?
- What do customers say about [your name]?
Write the answers down and save them. If they're thin, wrong or about a competitor, you know where to start. Ask the same questions again in a quarter. If there's something about you that you don't recognise, send it to us. We'll look at it together with you.