There's a ton of research showing that brand is an important metric for AI visibility and traditional search rankings. 

But how do you make your brand more visible in AI search when the old playbook no longer works?

I've spent the last couple of years experimenting on strategies that improve visibility for brands like Adobe, Marks & Spencer, and Clarks. 

In this guide, I'll share a repeatable blueprint you can use to win more citations and surface area in AI search.

But first, how did we get here?

The traditional search journey was messy and involved multiple stages, like:

• Exploration
• Evaluation
• Validation
• Conversion

A user would open multiple tabs, compare options, and check forums before making a decision. The journey was long and fragmented, requiring multiple touchpoints across platforms to convert.

AI search removes the friction. 

Instead of working through each stage manually, the user writes a prompt, and the LLM aggregates sources and condenses everything into a single answer. 

The messy middle still exists, but the user no longer navigates it themselves because AI handles it in the background and delivers a conclusion.

Search used to rely on short, high-volume keywords. For example, a query like “CRM” would sit at the top of the demand curve and attract a large share of searches.

As marketers, we could target those terms, build content clusters, and grow authority in a structured way.

Now, search is moving away from isolated keywords and towards conversational prompts. 

The search is no longer “CRM” but something much more detailed and specific to the user:

• A US-based SaaS company using HubSpot that needs better revenue recognition
• A field sales team that needs offline mobile access and location-based lead mapping
• A business managing complex territory mapping across regions and industries

These detailed queries create an infinite long tail.

At Journey Further, we analyzed over 100 client websites, 43 billion impressions, and 1.5 billion clicks to understand the changes.

Here’s what we found:

• Single and two-word queries are declining year-on-year
• Queries with four or more words are increasing significantly
• Five- to seven-word queries and eight-plus-word queries are growing fastest, and these tend to be comparison-driven and conversational.

As queries become longer and more specific, the number of possible variations expands, making it impossible to target a fixed set of keywords for AI visibility.

From what I’ve seen across AI search, four core signals determine visibility.

Entity clarity

AI needs to understand who you are and what you offer before it can recommend you. If your brand is ambiguous or inconsistently described across the web, the model struggles to place you in the right context.

To improve entity clarity:

• Build a logical site structure
• Use schema markup to define your brand, products, and services
• Keep your messaging consistent across channels

Technical SEO still plays a critical role by ensuring your brand is accessible and easy to interpret.

Authority and credibility

Authority and credibility come from how your brand is represented outside your site. AI models review what others say about you and use the information to decide whether you're worth citing. 

For example, in ChatGPT, 93% of citations are from third parties rather than your brand.

The signals they look for include:

• Third-party mentions in publications your audience trusts
• Reviews on platforms the model actively references
• PR coverage associating your brand with specific topics and categories

Positive, consistent signals increase your chances of earning brand mentions in LLMs.

Content completeness

AI pulls passages, not webpages. Structure your content so it's easy to extract an accurate answer. If information is difficult to retrieve, it is less likely to appear in LLMs.

That means:

• FAQs that directly answer the questions your audience is asking
• Guides that go deep on specific topics
• Answers formatted clearly beneath relevant headings

Resonance and presence

AI does not rely on a single source. The models look across the broader web to understand how your brand appears in conversations.

This includes:

• Reddit threads and forum discussions
• Social media conversations
• Earned media coverage

The more consistently your brand appears in the right places, the stronger your overall presence becomes.

Step 1: Map your audience’s search ecosystem

You need to understand where your audience spends time before trying to influence visibility.

Identify:

• Platforms they use
• Publishers they trust
• Communities they engage with

SparkToro is one of the most useful tools for this. It shows you where your audience spends time online, what they read, and where they engage.

For example, in the CRM space, we often see audiences who are heavily active on AI platforms like ChatGPT and Claude. It tells you where your brand needs to show up and shapes every decision that follows.

Step 2: Extract audience insight

Once you know where your audience is, the next step is understanding what they care about.

Reddit and its subreddits are among the richest sources of audience insight available. 

The conversations happening there reveal:

• Specific pain points your audience is trying to solve
• The language they use to describe their problems
• Questions they're asking before they make a decision

It’s important because Reddit is also one of the most cited sources in ChatGPT. The conversations your audience is having are the same ones AI is learning from.

Step 3: Build third-party validation as a core lever

Third-party validation is the most important lever you have for AI visibility. Here are a few PR tactics we’ve found successful at Journey Further:

Listicles and product reviews

Getting featured on “best of” lists and in recommendation content is one of the strongest signals of AI visibility. These formats appear consistently in LLM responses because they match how users ask questions.

Examples we see working for our clients include product features in recommendation articles like Delish and inclusion in curated guides like CNN’s “fiftieth birthday gifts under $200.”

The second placement is a perfect illustration of how this works. Someone types "can you recommend a fiftieth birthday gift under $200" into ChatGPT. The CNN piece surfaces, and Virgin Experience Gifts gets the citation.

The content doesn't need to be exclusively about your brand. It just needs to include your brand in the right context, on a publication the model trusts.

Data-backed research and consumer studies

Research-led content carries more weight than standard brand messaging because it provides the model with concrete evidence to reference.

We see this consistently with studies and data-led campaigns. For example, one of our clients, a Fortune 500 dog food brand, created a study on the most dog-friendly cities in America. Time Out picked up the research, and it’s now surfacing in AI Overviews for related searches.

Journalists are also prioritizing data-led sources to gain visibility in LLMs. When you pitch a study with great data and a credible source, you're giving them information they need to make their content citable in answer engines. 

Thought leadership and expert commentary

Thought leadership is one of the highest-impact ways to build visibility in AI search, and it is often underused.
 

Analysis from Journey Further's Salient tool across 4,000 pieces of DA40+ coverage found that nearly half included expert commentary or data studies rather than product content.

However, the demand for expertise has also created a problem where some agencies are fabricating expertise.

For example, in the UK, multiple publications picked up a story about a royal family cleaner before journalists discovered the expert didn't exist. 

In response, journalists now vet sources more rigorously than before.

There’s also reputational damage from using fake AI-generated experts, which has far-reaching effects on your brand.

Here’s what “good” looks like:

• Experts with credible experience
• Author bios published on-site
• Updated LinkedIn profiles
• Credentials that journalists can verify

Use digital PR as the distribution engine

Most teams still think about digital PR in terms of links, but that's too narrow.

Digital PR is the mechanism that makes everything else work, and fuels trust signals across three areas:

• Traditional search, where E-E-A-T still matters
• AI Overviews, where sources are selected and summarised
• LLM responses, where answers are built from multiple references

You either become the source that LLMs cite or ensure your brand is included in the sources they rely on for answers.

Step four: Follow a performance PR workflow

PR has a reputation for being hard to measure, but it doesn't have to be.

At Journey Further, every PR campaign starts with insight and ends with evidence. 

Here's the framework:

Insight and organic opportunity

Before any creative work begins, take time to understand where the opportunity exists, including:

• Keywords you rank for and those within striking distance
• Which industry publications show up more frequently in LLMs
• Where your audience spends time and what they're engaging with
• Gaps between your brand and competitors

This gives you a clear view of where you are visible, what’s missing, and where to focus.

Ideation and creative concept

Use the insight to develop concepts that are brand-led but audience-driven. The content needs to tell a consistent story about what your brand stands for and what it offers.

This is important because AI is pattern-matching across everything written about you. Hence, the more consistently your brand is associated with specific topics, the easier it is for the model to understand and recommend you.

Outreach

Target the publications you identified in the insight phase, including publications frequently cited by LLMs and websites your audience engages with.

Everything ties back to performance, so focus on platforms that influence AI outputs and audience decisions.

Further reading:

How to improve AI search visibility using what you already have

AI search platforms are walled gardens. There is no Search Console for ChatGPT or attribution model. But that doesn't mean you're flying blind.

You just need a different approach built around visibility, prompts, and citations.

Build a prompt bank based on your audience's wants and needs

Since you don’t have prompt-level volume data, the goal is to create a representative set of queries that reflect how your audience searches.

You can do it using three core sources:

• Google People Also Ask: A natural bridge between keyword and conversational query. It's not perfect, but it gives you questions with volume data as a proxy for demand.
• Reddit discussions: Shows the language your audience uses to describe their problems and ask for recommendations. These are also the conversations LLMs are trained on, which makes them valuable.
• AI search prompts: These are queries your audience is asking based on patterns from PAAs, forums, and your own research.

At Journey Further, the prompt bank we use for client tracking contains 2,000 questions and queries. They cover key intents across the full funnel, drawn from PAAs, FAQs, review sources, and forum conversations.

That bank becomes the foundation for everything else, including: 

• Tracking brand visibility
• Identifying citation gaps
• Measuring whether your PR and content activity moves the needle

Track brand visibility across multiple models

There are no rankings in AI search, so you need to measure visibility differently.

The metric that matters is brand occurrence, and here’s how to measure it:

• Track appearances across the models your audience uses
• Prioritize platforms based on your audience data
• Measure brand mentions across your prompt bank

This gives you a clear view of where you are visible and where you are not.

Next, go deeper into the cited URLs. Identify which sources the model uses when citing your brand or competitors. Identify responses where competitors appear instead and use the citation sources to determine where to focus PR efforts.

Use fan-out queries to understand what LLMs favor

Fan-out queries show how LLMs break a single prompt into multiple underlying searches.

For example, a query like “best CRM for an enterprise business” does not return a simple answer. The model expands it into different angles and synthesizes into an answer that addresses multiple versions of the query. 

When you analyze this at scale, patterns start to emerge. At Journey Further, we found:

• 41% of fan-out queries are nine words long
• 71% include superlatives like best, fastest, or cheapest
• Recency, pricing, and reviews appear consistently across queries

This tells you what LLMs prioritize when generating answers and provides direction for your strategy.

Track AI bot requests and correlate with performance metrics

The most reliable data you have is your own. Set up a bot request panel and correlate AI bot activity with the performance metrics you already track, including search console clicks, GA sessions, and conversions.

At Journey Further, we’ve seen a high correlation between search console clicks and ChatGPT requests at a page level. Content that performs well in traditional search is largely referenced by AI. Further proof that good SEO is good GEO.

However, the real value comes from the gaps. When a page is getting high search console clicks but low ChatGPT requests, it’s a sign that something is preventing AI bots from extracting the content. Find the gaps to improve on-site optimization.

The data also solves a problem most teams face when reporting upward. AI platforms are closed environments, so attribution is limited. Connecting bot data with your existing metrics bridges the gap and gives you something tangible to show stakeholders.

Visibility now depends on getting credibility across the web. Focus on becoming the trusted source through expert commentary,  data studies, and a PR strategy centered on credibility. 
There are no shortcuts to winning AI visibility. Brand mentions and citations are byproducts of a strong foundation and compound over time.