Improving AI visibility: seven changes for your business

Many people no longer start a search intending to visit a website. They ask a question and expect a direct answer, whether in ChatGPT, Perplexity, or the AI overview above Google's results. For purely informational questions, the search often ends there, with no click on a result.

That changes the goal for your business: in AI search, you aren't found; you're recommended. This article covers what changes for your visibility and the seven adjustments we make to ensure our clients are mentioned in AI answers. For the fundamentals, see our in-depth article on Generative Engine Optimization.

Fewer clicks, but visitors ready to act

Search clicks drop as soon as an AI summary appears. The Pew Research Center analyzed browsing data from 900 US adults in March 2025: with an AI summary, users clicked a result on 8% of searches; without one, on 15%.

AI acts as a pre-filter. Anyone who still clicks after getting an answer usually no longer just wants to learn but to act. Look at inquiries and purchases as well as clicks. That way you can tell whether fewer visits really mean less business.

1. We make your brand clearly recognizable to language models

For years, SEO was mostly about keywords: find the term with the highest search volume and optimize the copy for it. Language models work differently. They don't understand strings of characters; they understand things and concepts, known as entities: a brand, a person, a product.

If your brand isn't firmly tied to a subject area, whether in a model's training data or in the sources it searches when answering, it won't show up in the answer, no matter how often the keyword appears on your homepage. So we work to make sure a model “knows” who your company is, what it does, and what it stands for. We combine three things:

• Content with depth and original evidence.
• Visible expertise: who is behind the content, with author profiles and credentials.
• Reputation: what third parties say about your business.

When someone asks about leading industrial pump manufacturers, the model isn't looking for keywords. It's looking for a solid link between your brand and the topic.

How to check whether your brand is recognized as an entity

Search Google for your company name. If a Knowledge Panel appears, Google recognizes your company as an entity. If not, start by aligning your About us page, company profiles such as LinkedIn, and, where applicable, a Wikipedia entry so consistently that your identity is beyond doubt.

2. Your website gives machines direct access to all content

AI can only recommend what it can read. Google sets the same requirement for AI Overviews and AI Mode as for regular search: the page must be indexed and eligible to show a snippet. Beyond that, you need a logical site structure, clean internal linking, and an up-to-date XML sitemap that helps crawlers find your pages and understand how they connect.

JavaScript is a common bottleneck: many AI crawlers don't execute JavaScript, so pages that build their content in the browser look blank to them. Learn more in our article on AI crawlers and the JavaScript problem.

Structured data is still useful, but it isn't a lever for AI answers

Structured markup was long seen as the “language of AI”, and we, too, once described it as the most important lever for getting cited in AI answers. Google has since made it clear: AI search requires no structured data, and there is no special markup for it. Google Search doesn't use an llms.txt file either.

Structured data still belongs on every well-built website: it enables rich results and describes content unambiguously, such as your organization, articles with author and date, or products with price and availability. It must match the visible content.

3. Your content answers questions directly and backs up its claims

Content that answers a question up front serves readers and language models alike. So every section opens with the answer, followed by the reasoning and evidence. Comparisons go into tables, steps into numbered lists.

That doesn't mean breaking your content into fragments: Google explicitly says content doesn't need to be split into small chunks for AI search. What matters is something else. Google recommends content with a distinct perspective, experience, and expertise, and warns against simply repeating what others have already said. Generic knowledge that anyone can generate with AI gives a model no reason to cite you in particular.

Example: optimizing images for search

A weak paragraph opens with “In today's digital world, it is important for images to load quickly …” and takes several sentences to make a point.

A strong paragraph opens with the answer: “Optimize images for search in three steps: choose a modern format such as WebP, compress the file, and write alt text that describes what the image shows.” The details follow. But what gives the text real value is what only you know: data from your own projects, lessons from what didn't work, a well-reasoned recommendation.

4. We plan your visibility wherever your customers ask questions

Search hasn't meant just Google for a long time. Your customers ask questions on YouTube when they need a tutorial, on Pinterest when they want inspiration, on TikTok, and in AI assistants such as ChatGPT or Gemini. So we plan visibility across every place your customers ask questions.

For AI search, that means videos get carefully written titles, descriptions, and captions so their content is readable as text. Images and brand profiles on visual platforms are clearly labeled. According to Google, high-quality images and videos can also appear directly in Search's AI features.

So when we plan a client's visibility, the question is no longer just about ranking first, but whether they are the best answer, wherever the question is asked.

5. Your business gets recommended when others recommend it

AI has no opinion of its own about which product is best. It summarizes what credible sources say. If several trade publications describe a piece of software as the best accounting solution, the AI is very likely to mention it.

Traditional link building for rankings isn't enough. That's why we focus on digital PR: mentions in trade media, interviews, comparison articles, and best-of lists. For our clients, we identify which comparison and list articles exist in their industry and get cited in AI answers, then work specifically to get them featured. The path runs through media relations and relationships with editors, not paid placements.

How to assess where you stand

Search for “best [your industry] providers” or “top 10 [your product]” and ask an AI the same question. If you don't appear on any of the cited lists, the AI has no basis for recommending you when someone asks for buying advice.

6. Specific reviews make local businesses recommendable

For restaurants, service providers, and local retailers, AI search is a major opportunity. When someone asks for the best Italian restaurant in Munich-Schwabing for a quiet dinner for two, the answer draws on business profiles and reviews. Google explicitly names the Business Profile as a way to make offerings more visible in AI answers.

A review like “Super tasty, 5 stars” is nice to get, but it gives AI little to work with. A review like “The best tiramisu in Schwabing, perfect for a quiet evening for two in the courtyard” names the place, occasion, and product: exactly the details people ask about.

That's why we help our clients ask for meaningful reviews, for example with an open question in the follow-up email: “What did you like most about our [district] location?” Reviews are never incentivized, and requests are never limited to satisfied customers; both violate platform guidelines.

7. We measure how often AI answers mention your business

When the answer appears right in search, organic sessions alone say little about your visibility. So we've expanded our reports with three additional views:

• Visibility in Google Search: Search Console has a dedicated report showing how content performs in Search's AI features.
• Visitors from AI sources: We group visits from chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com as a separate segment in Google Analytics and report them separately for our clients.
• Mentions in answers: We use specialized tools and our own sampling to measure how often your brand appears in AI answers and which attributes the AI associates with it, such as “affordable” or “high-quality”.

A mention in an AI answer works like a third-party recommendation and is often worth more than an impression in position 3. So we count the mention itself, regardless of whether a click follows, and we factor brand awareness into the calculation we use to assess each channel's profit contribution.

Conclusion: AI search rewards businesses with real expertise

Adapting to AI search isn't a one-off technical trick. Average content and a handful of backlinks won't get you far in AI answers. Language models filter for quality: companies that show no real expertise, publish no original data, and neglect technical fundamentals rarely get mentioned.

For companies with genuine expertise in their field, this is an opportunity. Those who sharpen their positioning, publish their own experience and data, and get others to write about them will be the ones recommended in tomorrow's answers. To see how we approach this, visit our Generative Engine Optimization page.