- Jun 4
- 10 min read
Most conversations about generative engine optimization focus on how to get your brand into an AI-generated answer. But in order to do that, you need to understand that AI responses aren’t generic. It’s very rare that two people will get the exact same answer to a question.
Google's AI Mode, Gemini, and ChatGPT all use a mix of memory and personalization signals to shape what they surface to each individual user. The brands that show up in those personalized responses aren't always the ones with the highest domain authority. They're often the ones a user already has a relationship with: the newsletter they open every week, the YouTube channel they subscribe to, the loyalty program they check before booking a hotel.
In other words, repeat customer relationships are an AI visibility cheat code (as well as just good marketing). Here are some ways to optimize for personalization and build customer relationships.
01. Recognize that memory is not the same as personalization
Before we get into tactics, let’s be precise about what we mean by memory and personalization, because the distinction shapes the strategy.
AI memory is learned from patterns and habits
Memory in AI is inferred. AI systems observe user patterns like what a user searches for, what content they engage with, and what brands they interact with repeatedly, then build a model of that person's interests over time. You can't control what gets inferred.
This is why you can ask your favorite AI Tool, “based on our conversations, how would you describe my personality” and it will give you an overall summary of your tone of voice, typical approach, and interests. When I asked ChatGPT for mine, it gave me a fairly on-the-nose reading of how I tend to work.
When you use prompts to migrate memory between one platform, the export goes into much more detail, outlining brands and tools in preferences. So when I used Claude’s memory import prompt on my personal Perplexity account, many of the brands that I interact with daily—Wix, Semrush, Zapier, Google Sheets, VS Code—were listed in my preferences.
While brands can’t dictate exactly what ends up in the memory of a user, brands with strong customer relationships, ones that consistently create usable resources and rich experiences, are much more likely to be included here.
AI personalization is specified by the user
Personalization in AI is declared. Users actively configure their AI tools, setting their role, their preferences, their go-to platforms. These vary by platform but typically, personalization settings will ask questions about your profession and preferred tone of voice.
Personalization means that when two people ask the same question, they get two different responses based on their preferences. In the example below, two different lifestyle preferences returned different restaurants when I used the prompt “London brunch recommendations,” and the replies are more streamlined to help the user get to the best answer quickly.
A user who tells Gemini "I'm a freelance photographer who uses Adobe, Wix, and Lightroom daily" has handed that information directly to the AI. From that point forward, those brands have a built-in advantage in any relevant response.
This means you have two separate levers: build the kind of consistent engagement that infers well, and give customers reasons to declare their relationship with you. The tactics below cover both.
02. Use repeat visits as an algorithmic signal
We've all seen the "you visit often" tag that Google surfaces on sites you return to regularly. That's a signal that Google is paying attention to behavioral patterns. As AI Mode becomes more central to search, those patterns are more likely to shape the results users see.
Think about how social media algorithms work: if you regularly engage with a certain creator, you'll see more of their content. Google's Personalized Intelligence is moving in the same direction. Serving users "sticky" content they've already demonstrated interest makes good business sense for Google.
In practice, this means nurturing repeat visits is both a retention metric and an AI visibility strategy. Every return visit builds the behavioral signal that tells Google—and by extension, AI Mode and Gemini—that your brand is relevant and trusted by this specific user.
03. Optimize for memory with ongoing touchpoints and resources
Memory is inferred from patterns, which means the more touchpoints you create, the more pattern data you generate. But not all touchpoints are equal. The ones that matter most for AI memory are the ones where users actively engage with your brand in an AI context.
Demonstrable evidence is still emerging here, but we've seen a few formats that work well.
Create resources users that can pull into their AI tools directly. As AI assistants become more capable, users are increasingly saving, uploading, and referencing brand materials in their AI workflows. A well-structured knowledge base, a downloadable product guide in clean markdown, or a prompt template for your service category gives users something concrete to bring into their AI conversations. And every time they do, they're reinforcing your brand's association with their work.
Custom skills and GPT-like tools are another route. If your platform or product can be wrapped in a reusable AI skill or assistant, you're building a persistent presence in the user's AI environment rather than relying on one-off citations.
Write your knowledge base for both humans and AI. Clear, structured support documentation—with explicit questions, logical headings, and direct answers—is easier for AI systems to parse and cite accurately. The same article that answers a customer's question in your help center can also become the grounding source for an AI-generated answer when users want to unlock additional functionality in your tools via AI chat.
04. Optimize for personalization with interoperable tools
One of the most powerful declarations a user can make is connecting your product directly to their AI assistant.
If your platform integrates with AI tools directly, your customers can connect you to their AI environment. Wix, for example, is already available as an MCP (Model Context Protocol) connector in Claude and Perplexity Pro. That means a Wix user who connects their account can have their AI assistant work with live, up-to-date site content. No scraping. No generic responses. That kind of direct integration puts your brand inside the AI's working environment, which is qualitatively different from simply appearing in training data.
Prioritize integrations with the tools your customers already declare as part of their workflow. If you have an API or connector capability, make it as easy as possible for users to add your product to their AI toolbelt. The brands that live inside a user's AI setup have an enormous personalization advantage over the ones that don't.
05. Treat email as a grounding source
Email is underrated as an AI optimization channel, and I think it deserves a deeper look.
For Gmail users, research from Garrett Sussman at iPullRank Agency has shown that newsletters and the contents of a user's Gmail inbox can also influence AI Mode, when Personalized Intelligence is engaged. This adds additional AI search value to your mailing list.
Nurture your mailing list (and the content of your emails) with this in mind. Consistent email sends create behavioral patterns that AI personalization systems can observe. The user who opens your emails every Tuesday morning, clicks through to your blog, and replies occasionally has built a behavioral fingerprint that says "this brand is relevant to me." Serve them well in their inbox and you're building an AI signal at the same time.
In Garrett Sussman’s study, he found that the content of the emails had more effect on personalized AI mode responses than other highly personalized searches sources like photos.
This suggests that using consistent phrasing in emails builds context for AI. You can call them keywords if you like. The principle is the same: if you want to be associated with "sustainable activewear for runners" in an AI system, then "sustainable activewear for runners" should appear in your emails, not just your website. Think in semantic associations. As in, not just the product name, but the category, the audience, the use case, and the occasion. Make those associations explicit and consistent across every email you send.
And don't overlook your footer and boilerplate. The email footer is one of the most consistent, high-frequency pieces of copy your brand produces. If it includes your category, your audience, your key differentiators, and your brand name, then every single email reinforces those associations.
06. Build YouTube and social into your full marketing machine
YouTube receives millions of visits from AI chatbots every month. AI systems are actively learning from the content on YouTube but also from user behavior.
YouTube, in particular, is part of Google's Personalized Intelligence grounding sources. A user who regularly watches your channel has created a behavioral signal that Google can draw on when that user asks a relevant question in AI Mode and Gemini. A user who has your channel linked in their Gemini settings is even more explicitly connected. You can leverage this engagement for personalization optimization.
Consider all your YouTube touch points. Think about text (titles, descriptions, transcripts, chapters), images (thumbnails, screenshots), structured metadata (tags, categories, playlists), and community posts. Each of those layers gives AI systems more signals to work with—and gives your brand more surface area in the personalization layer.
That’s why I suggest treating YouTube as an engagement hub, not just a video platform. Write full descriptions with the same semantic associations you'd use on your website. Use chapters and timestamps to create structured content that AI can parse. Create community posts that mirror your email content, so users who engage with your brand across multiple surfaces are reinforcing the same associations.
I’m not recommending that you make YouTube your entire strategy, but be intentional. Integrate your YouTube activity strategically with your broader marketing engine; without these deliberate links, personalized AI systems are left guessing.
07. Use brand activations and visual search to build affinity at scale
This one is less obvious, but it’s worth thinking about as visual search capabilities continue to expand.
Google Photos is connected to Gemini's Personalized Intelligence. That means photos stored in a user's Google account—including photos taken at events, photos saved from social media, and photos that feature branded merchandise or environments—can become part of the context the AI uses to understand that user's world.
When connected to Google Photos, Gemini can analyze images to make recommendations. So when I asked, “based on my photo[s], where do I have the most fun on vacation,” it accurately said that I have “a great time at the beach and ocean.”
It can also pick up brands in photos. So when I asked “based on my photos which brands do I engage with most?”, unsurprisingly Wix came out on top.
This means that potentially, if a user regularly wears your branded merchandise, attends your events, or saves images associated with your brand, those visual associations can strengthen your presence in their personalized AI environment. Think about how this works in practice. A user who owns a team jersey and has hundreds of match-day photos in Google Photos will likely see that team featured prominently when they ask their AI assistant about sports tickets or merchandise.
Brands can build deliberately for this.
Create events and activations that give customers a reason to photograph and share branded moments.
Optimize for visual search by ensuring your logo is placed prominently and consistently: on merchandise, event materials, packaging, and in-store displays.
Work with your brand team to validate your logo as a recognizable entity in visual search tools.
Create "instagrammable" moments that customers will want to photograph and keep. The more those images exist in your customers' photo libraries, the more consistently your brand appears in the visual layer of their personalized experience.
08. Manage reviews and communities as personalization signals
On-site reviews are content. The more specific they are, the more useful they are as personalization signals. A review that identifies the reviewer's age, activity type, and use case (as Gymshark does, requiring reviewers to specify their workout frequency and fitness goals) isn't just social proof for prospective customers. It's audience signal data that tells AI systems exactly who your product is for, in the language of your actual users.
Add contextual clues for your reviewers. Make it easy for customers to identify themselves in reviews, not just a star rating and a comment, but a few data points about who they are and how they use your product. Those data points become the vocabulary that AI uses when matching your product to a relevant user.
Off-site, engage consistently with customer communities. Reddit, niche forums, and brand communities are places where your ICP already gathers and talks about their needs. Showing up there (with genuinely useful contributions, not marketing copy) builds the kind of third-party citation pattern that AI systems use to understand a brand's relevance in a specific category.
09. Build long-term influencer and community partnerships
Influencers are remarkably effective in AI search: consistent, multi-channel brand presence in the feeds of a specific audience.
When a brand is mentioned consistently across an influencer's YouTube videos, their newsletter, their special offers section, and their social posts, that brand becomes part of the information environment for every subscriber and follower. Those subscribers and followers are generating behavioral data that AI personalization systems can draw on.
The key word is long-term. A one-off sponsored post creates a single data point. A 12-month ambassador relationship creates a pattern, and patterns are what AI memory systems are built to detect. Invest in partnerships where your brand becomes genuinely woven into a creator's content and community, not just an occasional sponsored mention.
10. Love your loyalty programs
Loyalty programs offer a mutually beneficial route for brands to engage and reward their most committed customers. In an AI-personalized environment, they do something else, too: create explicit, declared brand associations that users bring directly into their AI interactions.
Think about how often people include loyalty considerations in their purchasing decisions. "Only if I get Hilton Honors points." "Do they offer a student discount?" "Is there a military rate?" These are things people actively state when planning purchases, including when they're using AI assistants to help them plan.
A user who asks ChatGPT to help them book a hotel and specifies "I need it to earn Hilton Honors points" has just declared a brand preference directly to their LLM of choice. These types of declared associations are exactly what loyalty programs and segment-specific offers create.
Create dedicated programs for your most important customer segments. When Wix promotes a "Wix for Students" program on Student Beans, it's building a named association between its brand and a specific audience segment, one that both students and AI systems can recognize and reference. Show your loyal customers some love. It's good for them, it's good for your brand, and it turns out it's very good for AI visibility, too.
The same goes for the other tactics on this list. Relationships that exist in the real world, on real channels, with real customers, create the signal patterns that AI personalization systems are designed to surface. You don't need to choose between good marketing and AI optimization. In a personalized AI environment, they're the same thing.
