- Answer Engine Optimization (AEO) starts with measurement: 20 questions your customers ask, tested every month in ChatGPT, Perplexity and Gemini.
- Google says optimizing for AI search is still SEO. Clean technology, clear structure and indexable pages remain the foundation; Google doesn't need an llms.txt file.
- Content with numbers, quotes and cited sources is picked up more often in AI answers; one study measured up to 40% more visibility.
- Merchants supply their product data directly: to ChatGPT via the Agentic Commerce Protocol, to Google via Merchant Center.
More and more buying decisions start with a question to ChatGPT, Perplexity, Gemini or Google's AI Mode. The answer names only a few companies and products. If you're not among them, you're invisible to part of your future customer base.
Answer Engine Optimization (AEO) ensures that AI systems select and recommend your company as a source. In its July 2026 guide, Google makes clear that optimizing for AI search is still SEO: there's no need for special files such as llms.txt, special structured data or breaking content into small chunks. Our 24-point checklist shows where to start, from measurement, technology and content to mentions on other websites.
First, we measure how AI sees your brand today
1. Measure your selection rate
What it's about: Regularly measuring whether your brand appears in AI answers to relevant questions. This gives you a solid basis for assessing your presence objectively.
Test your 20 most important questions manually. Don't just record whether you appear, but also which competitors the AI names instead. That shows you exactly who has digital presence in your niche.
Alex Sperber, Partner at UnitedAds
How to do it: Create a spreadsheet with 20 questions that define your category. Test them every month in ChatGPT, Perplexity and Gemini and record: mentioned or not, in what position and in what tone.
Why it matters: Without a baseline, you can't prove your work is paying off.
Goal: a measurably higher mention rate and better positions in AI answers.
2. Check how AI perceives your brand
What it's about: Examining what “opinion” AI models hold about your brand and industry, and where their knowledge falls short.
AI models don't just hallucinate; they have biases based on old internet data. If AI assistants consider your offer expensive, they won't recommend it as the best value, no matter what your pricing page says.
Alex Sperber, Partner at UnitedAds
How to do it: Ask the models directly: “Which brands are known for [category]?” and “What sets [brand] apart?”. Document how they describe you and close the gaps with targeted content on your website and on third-party sites.
Why it matters: The models rely on existing content on the web. Incorrect or outdated descriptions only disappear if you actively counter them with accurate information.
Goal: close knowledge gaps and sharpen your brand's profile in AI answers.
We make sure crawlers can read your pages quickly and completely
3. Reduce server response time
What it's about: Speeding up the Time to First Byte so crawlers can fetch your pages quickly.
AI bots are less patient than human users. If your server is slow, the crawler gives up on the extraction and takes the data from your faster competitor instead.
Alex Sperber, Partner at UnitedAds
How to do it: Measure the TTFB of your 20 most important pages with tools such as WebPageTest. Use a CDN and server-side caching, and optimize database queries.
Why it matters: Fast pages are crawled more reliably, and your visitors benefit too.
Goal: a fast, consistent server response on all important pages.
4. Put the main keyword at the start of the page title
What it's about: Placing the central topic of a page at the start of its title.
How to do it: Revise the titles of your 50 most important pages using this pattern: [Main topic] | [Secondary topic] | [Brand].
Why it matters: The title is a page's shortest summary. With the topic up front, search engines, AI systems and readers grasp it instantly, even if the title is truncated.
Goal: each page clearly matched to the questions it answers.
5. Write meta descriptions as factual summaries
What it's about: Meta descriptions that summarize the page's content factually instead of simply asking for a click.
How to do it: Write each description as a brief summary: the main topic, the key facts and who the page is for, in about 140 to 155 characters.
Why it matters: A clear summary helps readers decide in search results and ensures your page is understood correctly.
Goal: accurate understanding of your content and more relevant clicks.
6. Simplify the URL structure
What it's about: Creating a clean, readable URL hierarchy without cryptic parameters.
How to do it: Remove session IDs and tracking parameters from internal links. Use hyphens instead of underscores. Leave years out of URLs for content you update regularly so the address stays valid long term.
Why it matters: Clear URLs reveal a page's hierarchy and topic in the path itself and stay stable when content is updated.
Goal: clean indexing and a clear understanding of your topics.
7. Supply product data directly to AI providers
What it's about: Merchants supplying their product data directly to AI providers' shopping features instead of waiting to be crawled.
How to do it: For ChatGPT, merchants share product feeds and promotions via the Agentic Commerce Protocol, directly or through providers such as Salesforce and Stripe. Shopify merchants are already connected via Shopify Catalog with no effort on their part. For Google, Merchant Center remains the route for getting products into AI answers as well. Complete all required fields such as title, price and availability, update daily and review error messages.
Why it matters: Data supplied directly is more complete and current than crawled data. That improves your chances of being recommended when people ask for buying advice.
Goal: presence in AI shopping recommendations, with accurate prices and stock levels. Learn more in our article on agentic commerce with Shopify.
8. Use semantic HTML
What it's about: Marking up the meaning of your content with the right HTML elements.
How to do it: One H1 per page, with sections below it using H2 and H3. Use <table> for data, <ul> for lists and <ol> for processes. Make sure the content is in the HTML without relying on JavaScript, because many AI crawlers don't execute JavaScript.
Why it matters: These elements help machines distinguish content from design and navigation.
Goal: accurate extraction of facts, tables and lists.
We turn your content into a citable source
9. Increase fact density
What it's about: Replacing generic phrasing with concrete numbers, data and names.
How to do it: Check every paragraph: Does it open with a clear statement, and is that statement backed by a number, an example or a source?
Why it matters: The study that coined the term GEO (Aggarwal et al., KDD 2024) found that targeted changes to a text increase its visibility in AI answers by up to 40%. Statistics, quotes and cited sources had the strongest effect.
Goal: being selected as a reliable source for AI summaries.
10. Collect your customers' questions
What it's about: Building a collection of real customer questions as the foundation for your content.
How to do it: Analyze support tickets, sales conversations and forum discussions. Organize them by question, answer, source and priority.
Why it matters: People ask AI the same questions they ask your sales team. If you know and answer those questions, you give the AI exactly the material it needs.
Goal: your answers get picked up in AI responses. Group related questions on strong pages instead of creating a separate page for each one; our article on the risks of AI content explains why.
11. Phrase questions in natural language
What it's about: FAQs that sound the way people actually ask.
How to do it: Write questions using question words (how, what, why) and read them aloud. Rephrase anything that sounds unnatural when spoken.
Why it matters: The closer your wording is to the user's question, the easier it is for the AI to match your answer to it.
Goal: the closest possible match between user intent and your content.
12. Update important content every quarter
What it's about: Introducing a fixed process for regularly updating your most important content.
How to do it: Review your 20 most important pages every three months. Update numbers, examples, FAQs and links. Change the date only if the content has actually changed.
Why it matters: AI answers rely on up-to-date pages. Outdated numbers undermine your credibility as a source.
Goal: no outdated information in AI answers about your company.
13. Display the date of the last update
What it's about: Letting readers and machines see when the content was last revised.
How to do it: Display the date prominently below the title and use the HTML element <time>. The date in the text, the sitemap and the structured data should match.
Why it matters: A visible date shows readers and systems how current the information is.
Goal: more reader trust and clear freshness signals.
We build your credibility across many sources
14. Earn links with descriptive anchor text
What it's about: Getting topically relevant sites to link to your content with descriptive anchor text.
How to do it: Create content others want to reference, such as original research or in-depth guides, and pitch it to respected industry sites. Aim for descriptive anchor text rather than “click here”.
Why it matters: Links from industry sites confirm what your brand stands for. Google points out that fake mentions don't work, because AI search relies on the same spam systems as regular search.
Goal: stronger authority and a sharper brand profile in your subject area.
15. Publish joint content with partners
What it's about: Partnering with other brands so your names are mentioned together.
How to do it: Team up with a few brands whose offering complements yours on guest articles or joint studies, with each brand mentioned in the text.
Why it matters: Language models associate brands that are frequently mentioned together. Being linked with established brands in your industry strengthens your own positioning.
Goal: a firm position for your brand in your subject area.
16. Maintain profiles on review platforms
What it's about: Maintaining your presence on review sites such as G2 or Trustpilot.
How to do it: Verify your profiles, complete every field, add images and respond to reviews within 48 hours.
Why it matters: AI answers often draw on review sites as independent sources for assessing providers.
Goal: a consistently positive picture in the sources the AI relies on.
17. Structure internal linking by topic
What it's about: Linking your pages in a way that makes topic clusters and expertise visible.
How to do it: Find pages ranking in positions 11 to 30 on Google and link to them from strong overview pages on the same topic.
Why it matters: This lets search engines and AI systems grasp the full depth of your knowledge on a topic.
Goal: authority across entire subject areas, not just individual pages.
18. Answer comparison questions and complex questions
What it's about: Explaining comparisons and relationships between products, providers and concepts.
How to do it: Create content for questions such as “[Brand A] or [Brand B]?” or “How does [Product A] work with [System B]?”, based on real-world use and with transparent criteria.
Why it matters: In AI search, people ask for advice and comparisons. Your content needs to answer these questions honestly.
Goal: becoming the AI's main source for advice and comparison queries.
19. Name authors and their credentials
What it's about: Showing on every article who wrote it and why that person knows the subject.
How to do it: Place the name, role and photo at the top, add a short bio and link to an author page and LinkedIn.
Why it matters: Experience and expertise (E-E-A-T) are criteria Google uses to evaluate content. They also make it clear to readers who stands behind a statement.
Goal: proof that the content comes from a qualified person.
20. Display trust signals prominently
What it's about: Clearly showing certifications, awards and results.
How to do it: Display certifications such as ISO or SOC 2 and industry awards prominently, and link to the corresponding verification page.
Why it matters: Verifiable proof can be extracted and cited. It gives readers and AI systems a reason to trust you.
Place relevant evidence right next to the claims it supports, such as customer reviews, qualifications or verifiable project results.
21. Replace claims with evidence
What it's about: Replacing marketing clichés with verifiable facts.
How to do it: “Leading” becomes “No. 1 in Report X”. Every marketing claim becomes a verifiable piece of information.
Why it matters: Unsubstantiated self-praise adds nothing to an answer. Facts are what get cited.
Goal: more credibility and statements the AI can use.
22. Take part in discussions on platforms
What it's about: Monitoring and taking part in conversations about your brand on Reddit, YouTube and industry forums.
How to do it: Set up alerts for brand mentions. Respond to comments, offer help and thank people for positive feedback.
Why it matters: AI answers frequently cite forums and platforms. What people say about you there shapes the picture the AI paints of you.
Goal: a positive foundation for how your brand is presented in AI answers.
23. Add transcripts to videos
What it's about: Providing an accurate transcript on your website for every video.
How to do it: Create videos on common customer questions. Review and correct the auto-generated transcripts and embed the text directly on the page.
Why it matters: Text is easier to extract than video. Transcripts make the knowledge in your videos accessible to AI search.
Goal: your video content makes its way into AI answers.
24. Show up on Reddit with help, not advertising
What it's about: Building a reputation in Reddit communities through helpful, non-promotional posts.
How to do it: Be active in a few subreddits related to your topic. Rule of thumb: ten helpful posts for every one about your own business.
Why it matters: Reddit discussions frequently appear in search results and AI answers. What gets recommended there shapes the answer.
Goal: appearing as an unprompted recommendation in the AI's sources.
Conclusion: AI answers recommend brands that can prove their substance
Most items on this list are good SEO, and that's exactly what Google says too. What's new is the goal: instead of a position in the results list, what counts is being named in the answer. That takes a baseline measurement, technically clean pages, content backed by your own data and evidence, and a brand that others write about too.
A plugin or a handful of generated texts won't get you there. When everyone writes with AI, the sources that get selected are the ones that contribute something original. We describe our approach in our article on Generative Engine Optimization.
