Lea is in front of her laptop looking for sports cars. So, her AI assistant helps her selecting the perfect car by giving her a short list of vehicles that fit all of her criteria for price, horsepower, fuel and more. Lea can even read a review for each of the cars on the list that her AI assistant has put together for her.
Back on the webpage comparing all the different sports cars Lea had looked at previously she is able to fully review all of them again. So, one of the cars has excellent design, the other lots of great history and finally there is a very practical car that Lea feels she should buy, however Lea is having difficulty picturing herself driving this car.
That’s the awkward bit about AI visibility. A brand can be named, compared, or even recommended and still be misunderstood. For some companies, that is a small price to pay for being easier to find. For others, a short answer can leave out the very thing customers value most.
What does it mean to appear in an AI answer?
In order to manage a brand’s image in these new search experiences, however, it is not a matter of trying to get a mention in a short summary of a search query, rather it is largely a function of the brand’s online context and customers’ expectations of it. This article outlines how a number of current generative search features and answers have been generated and also goes on to develop this in more detail in order to consider the consequences for brands of current generative search features and how they might be managed.
The term Generative engine optimization (GEO) has in recent times been used by researchers and practitioners in order to describe the way in which companies can maximize the information that is published about them within generative search engine results (Aggarwal et al., 2024). In the light of Google Search’s new AI-powered search engine features however, the distinction between GEO and SEO (search engine optimization) for example is not as clear-cut as has been claimed in the past, since all of the Google Search features mentioned are powered by the Google Search index, and therefore employ the same fundamental SEO practices that apply to Search (Google Search Central, 2026).
When the tidy answer misses the point
Most simple searches for everyday products and services can be efficiently answered in a short list (e.g. for a compatible phone charger, local repair shops etc.). There is no need for lengthy brand descriptions.
But, a luxury car like the Ferrari example above can perhaps be summarized by comparing in short form other competing cars by price and features but much of what is perceived by the consumer to make ‘luxury’ is about perceived rarity and exclusivity of the luxury brand (Wang et al., 2022).
An owner of a small business offering private stays on an island for example wants her company to be listed in search results for “Where can I book an island vacation?” Of course there are many other vacation packages for island destinations and some of them might be chosen by users searching for such travel in short form answers. But other users reading the answer would associate the company offering such an answer with typical packaged vacations and thus would not be interested in the unique experiences offered by this owner of a small business.
What more?
Many booking inquiries consist of more than just an offer to book. Introductions and many questions are the norm. If one wants to book well with a customer, he or she must feel understood. Thus a trusted referral is much more valuable than high rankings.
There are several reasons for which it might be undesirable for confidential information of a company to be included in search results. Such information includes launch dates, customer lists, unsold designs, as well as current promotions. Google recommends to keep confidential content (e.g. for holiday bookings) password protected. On top of the company’s own web pages also the robots.txt file of a website can be used to declare which parts of a website search engines should and should not index. However, for confidential information that is published on the web on the web, access controls are by far the better way to control visibility of such content.
A mention is not a recommendation
As noted earlier, an AI answer can contain a reference to Lea, such as a description of a product they offer and even a recommendation for said product and link to the relevant web page. This mention or link does not necessarily equal to an endorsement by the AI, nor does it equal to clicks on said link by the users of said AI.
Even when AI search tools consult current webpages for their answers, familiar SEO practices remain relevant for Search, including for how the brand is represented in its answers (Google Search Central, 2026). Just because your information has been resulted for recently, does not mean it has been visited recently by users. In online research of browsing history for 900 U.S. adults over time, Pew found that in instances when Google generated a brief AI summary for a search query, 8% of subsequent clicks on Search results led to a blue link compared to 15% when there was no summary provided by Google. However, only 1% of all clicks within the summary’s link detection powered by Google were of links embedded within the generated summary (Pew Research Center, 2025).
So, should your brand show up?
When considering the search query above, we would first ask who is asking the question and what are they looking for. If someone is trying to compare a couple of products then yes, make it easy for them to read your Product Features and Prices. But if someone is trying to understand a complex offer then try to contain your Product Features and Prices to the bare minimum and point them in the direction of where they can read a more in depth description of your service from a human.
| What the customer needs | A possible approach | Result / Trade-off |
| A quick comparison or nearby service | Appear broadly | -More people may find the brand -Short answers can lose important context |
| Help understanding a complex offer | Appear selectively | -Useful facts stay visible while custom or private details stay controlled |
| A private or invitation-only experience | Limit visibility in some AI features | -Fewer broad impressions -Other sites and AI tools may still mention the brand |
Wait a minute! Before you start changing things around, it is a good idea to first clearly outline what your company is selling, who can buy it and what the limits of your service or offer are. Then use the AI tools for long-tail searches for questions that your customers would normally ask, save the answers, include the date you saved them and indicate the sources that you used. Note any errors in a simple spreadsheet.
However, just knowing a brand has been mentioned is not enough. The true value of a mention for a business will depend on whether the customer that the mention was intended for actually reads the content of the mention and understands the key differentiators of the brand’s offer.
What can you do?
The level of control offered to brands to control how they appear in AI search varies by platform. Google Search generative AI control is offered within Search Console for Site owners to include / exclude their site from AI Overviews and AI Mode. Note that this only controls Google’s Search generative AI and does not control other AI tools. Furthermore, being removed from Google Search AI answers does not mean that the site has been removed from Google Search entirely (Google Search Central, 2026b). OpenAI’s OAI-SearchBot is used for ChatGPT searches and can be excluded by brands in OpenAI settings, whereas GPTBot is used to crawl content so that it can be used for training of models and cannot be excluded by brands (OpenAI, n.d.).
And finally, one last thought: even an odd and wrong answer needs to be followed by you. You trace back where you got that from, was it on your site or on some external page that’s become outdated? Update that as necessary and use the feedback options provided by the AI platform in question. As with all your visibility efforts, one wrong answer to an odd question doesn’t merit scrapping the whole approach.
A few questions that come up
How can a brand keep private information out of AI answers?
Confidential information should only be published on a company’s own website, and behind login details where appropriate. A robot.txt file is not a privacy control and even if you have restricted information on your website, copies of and details of that information could still have been published by other sources.
And No. A mention or link does not mean the company approved the answer. Make the company’s own policies easy to find and correct important errors at their source.
Should luxury brands avoid AI search altogether?
Product-care information or approved retailers are generally good to show widely. But when it comes to custom services or private access to a unique product or experience, this often needs to be offered by invitation. In such cases visibility in AI-based discovery is less relevant and can even cause more harm than good.
When should a brand opt out of AI search?
Whether to limit a search engine’s crawling of material on your site in order to prevent sensitive information or an erroneous characterization of your service(s) being ‘discovered’ by potential customers is a decision to be taken on a case by case basis.
Being seen is only the beginning
However, when Lea gets to read about all cars of your brand, she can choose. An AI comparison can only help her getting oriented. But what will she care about after the first test drive? So, Will it be the car that looked best on paper or the one she can’t get out of her mind while driving home?
Ultimately, being seen is just the starting point for a brand. While information being easily available is helpful for a brand to be considered for purchase by potential customers, it is only the first step in getting them to choose your offering over that of another brand. Where information is provided, it must also be used.
References
- Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., & Deshpande, A. (2024). GEO: Generative Engine Optimization. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. https://doi.org/10.1145/3637528.3671900
- Google Search Central. (2026a). Optimizing your website for generative AI features on Google Search.https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Google Search Central. (2026b). Control what you share with Google Search.https://developers.google.com/search/docs/crawling-indexing/control-what-you-share
- OpenAI. (n.d.). Overview of OpenAI crawlers. https://developers.openai.com/api/docs/bots
- Pew Research Center. (2025, July 21). Do people click on links in Google AI summaries?https://www.pewresearch.org/short-reads/2025/07/21/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- Stanford Human-Centered Artificial Intelligence. (n.d.). What is a large language model (LLM)?https://hai.stanford.edu/ai-definitions/what-is-a-llm
- Wang, Y., et al. (2022). Examining the influences of perceived exclusivity and perceived rarity on consumers’ perception of luxury. Journal of Fashion Marketing and Management, 26(2), 365–382. https://doi.org/10.1108/JFMM-12-2020-0254