The AI Search Optimization Guide for LLM Visibility

AI search is going to replace Google's blue links. The future is AI search optimization. Here's the breakdown.

Online marketing is fundamentally changing. Google search traffic has been the bedrock of most online strategies. That traffic is now disappearing into AI chat.

In the following post, I will break down what we know to be true about AI search and how to approach AI search optimization to increase visibility in LLMs.

As always, I've created a fun prompt to help you with AI search optimization and tried it out with the newly released o3 Pro - a video is included at the bottom of the post, along with the prompt.

P.S.:- I'll be posting a tutorial on how to create a prompt engineer for ChatGPT 4.1 models next week for my paid subscribers. Sign-up to get the video tutorial

What do we know?

1. AI assistants are growing: ChatGPT allegedly just passed 1B users in April, and Gemini has 400M+ users. We're seeing mass market adoption of these tools.

2. For the first time in 22 years, Apple reported that Google Searches on Safari decreased in May.

3. Zero clicks are the norm: a Datos × SparkToro study of 332 million queries found that 58.5% of US Google searches and 59.7% in the EU end with no click at all. Only 360 of every 1,000 US searches send a visit to the open web (374 in the EU). sparktoro.com

4. AI Overviews are on the rise - they appear in 54% of Google searches by volume. And they suck away clicks: Ahrefs' 300-keyword study shows a 34.5% drop in CTR for the #1 organic result when an AI Overview is present. Google has alluded that AI mode will become the default experience for searchers in the US.

5. AI search visitors will surpass traditional search visitors by 2028 (imagine just 18 months ago saying Google’s traditional search engine would be replaced in under 5 years!)

6. The average LLM visitor is worth 4.4x the average traditional organic search visitor. By the time visitors from AI search arrive on your site they’re ready to buy.

Takeaway: There is no going back. AI search will be the predominant way people research and buy products in the future.

SEO -> Brand Marketing

On LinkedIn, I recently said that:

'SEO isn't dying. It's evolving into TV ads. Everyone sees your Ad, and no one clicks it. AI assistants turn search into a billboard.'

SEO was great because you could track the return on your investment. You could figure out the keywords relevant for your product and exclusively target those.

AI search is different. People don’t click. People don’t use keywords, they speak with their AI assistant like a friend.

At HubSpot, we've seen a 500% increase in referral traffic from ChatGPT in 2025; however, it still accounts for a relatively small portion of traffic.

However, there is still a real opportunity to surface your products in AI search for relevant questions. The people who learn to do this first will reap the rewards. AI search will be the best way to gain visibility for your brand. It will look more like 'brand marketing' than 'performance marketing.'

For example, here's one of the charts we obsess over from xFunnel - 'share of voice,' which shows how visible we are for the questions relevant to our product.

What should you do?

Ok, so if AI search is so new and most AI researchers will tell you they don't know why AI gives the answers it does, is there anything you can do to influence visibility for your product?

Yes! I’ll focus on two of the biggest.

1. Focus on the long tail: AI assistants don't rely on a single broad keyword. They field thousands of ultra-specific questions, and the pages that answer those questions win the citation.

The long tail is the thousands of ultra-specific questions people ask about your product. Each might have a tiny search volume, but with AI search growing so rapidly, all these small searches will collectively add up to significant visibility for your products.

Why does this work? LLMs lift snippets, not pages. They quote the first concise answer they trust. Long-tail questions will face little competition. Brands have traditionally focused on more focused keywords. It will take time for most brands to pivot from this strategy.

How do you go after the long tail?

a) Spin up role-, industry-, and use-case variations

People don't just ask, "Is HubSpot CRM good?" They ask, "Is HubSpot CRM good for a small manufacturing company?" or "How can HubSpot help a sales leader grow more revenue for their business?"

You should:

• Create one dedicated page or section for each high-value slice—roles, industries, regions, or specific “jobs to be done.”

• Keep about 70% of the copy shared (your core pitch) and swap in 30% of tailored examples, benefits, and FAQs so that Google and LLMs see each page as genuinely different. Obviously this may different depending on how unique you can make the content.

b) Surface unique, proprietary data

The long tail is much easier when you cite numbers no one else has. Every business will need to determine its unique data set to create variations of pages, such as benchmarks, cost savings, industry case studies, usage patterns, and more. There’s a case to be made that you’ll eventually have product pages tailored for individual companies. I call this precision marketing. More on that in future posts.

2. Earn Credible Citations

It's no longer about links; citations/mentions are your new link-building strategy.

The challenge is each AI search engine has its own preference for sources it uses for citations. These citations follow closely with websites the companies have partnerships with.

All LLMs are regulary citing sources as part of their results that are outside of the top 20 of Google’s search results, with ChatGPT citing these sources 90% of the time. It highlights that AI search is different in how it ranks URLs than Google’s traditional search engine.

How do you go after credible citations?

• SEO still matters: Google AI overviews still favor URLs already in the top results. You still need to care about your keyword placement for transactional keywords e.g. those keywords people use when in buying mode.

• Unique data matters: If you have proprietary data you can turn it into a press-worthy mini-study. Pitch it to respected trade journals or Tier-1 news outlets; Google’s model disproportionately amplifies those domains. (Remember, the BBC and NYT dominate 80 % of news citations in AI Overviews). For ChatGPT you’ll need to target a different set of media sites with those stats.

• Community-drive websites: Reddit and Quora are two of the most cited sources in Google AI overviews. A study released by Ahrefs showed Reddit was in 7.4% of all answers (third overall, after Wikipedia and YouTube) and Quora was 3.6% making both sites the only user-generated forums that crack Google’s top 10 citation list. Google AI overviews is also pulling in data from YouTube and LinkedIn (see image above). Where as OpenAI is leaning heavily on Reddit. Brands will need to have strategies to building into those communities - YouTube channel, AMA’s on Reddit, Q/A strategy on Quora, thoughtful content on LinkedIn.

There’s a lot more advice on how to optimise your website/pages for AI search but the above are, I believe, two of the biggest areas to figure out.

Here’s a fun prompt based on the core learnings I put together for this post to help you do AI search optimisation for your product. I tried it out using o3-pro, results in the video:

---

PROMPT

# Role and Objective

You are a strategy assistant trained on the most effective, research-backed methods for improving product visibility in AI-driven search results, including Google AI Overviews and ChatGPT answers.

Your task is to take a product description from the user and return a custom LLM optimization plan that helps the product become more discoverable, quotable, and indexable by modern language models and AI crawlers.

# Instructions

- Ask the user to briefly describe their product (what it does, who it helps, what type of content supports it).

- Then use the **LLM Visibility Optimization Checklist** below to build a tailored plan for their product.

- Organize your plan into five sections:

1. Page Structure for Query Fan-Out

2. Trust Signal Enhancements

3. Crawler Accessibility

4. Monitoring & Iteration

5. Why This Matters

- Do not make up strategies. Only use the ones listed below.

- If the product is unclear, ask a clarifying question.

---

## 📋 LLM Visibility Optimization Checklist

### 1. Page Structure for Query Fan-Out

These tactics help your product content answer multiple sub-questions that LLMs decompose from a user query:

- Add 6–10 H2s per page, each answering a likely sub-question about the main topic.

- Begin the page with a ≤150-character TL;DR followed by a 3-row stat table with key facts or numbers.

- Present important data in multiple structured formats:

- HTML table

- JSON-LD using `Dataset` with `variableMeasured`

- Downloadable CSV or JSON file

### 2. Trust Signal Enhancements

These help your content be selected and cited more often:

- Publish core product stats or insights on high-authority third-party platforms (e.g. Wikipedia, Reddit AMAs, analyst newsletters).

- Clearly state cohort size, methodology, and last updated date in both visible copy and schema markup.

### 3. Crawler Accessibility

These practices make your product content easier for AI crawlers to parse:

- Server-render all key content; use JavaScript only for progressive enhancements.

- For data visualizations, include a static PNG fallback inside a `<picture>` element, with descriptive `alt` text.

- Edge-cache your pages and aim for a Time to First Byte (TTFB) under 100 ms.

### 4. Monitoring & Iteration

These steps ensure ongoing visibility:

- Run daily self-queries for your key H2 topics and log whether your product is cited.

- Monitor whether AI Overviews appear for your target queries and track drops in click-through rate (CTR), which may indicate AI displacement.

### 5. Why This Matters (For Exec Buy-In)

- ChatGPT attracts over 5.5 billion visits per month — it’s the 6th most visited site globally.

- AI Overviews now appear in 13.14% of U.S. search results.

- These AI results reduce position-1 CTR by 34.5% and contribute to the 58.5% of U.S. searches that end with zero clicks.

---

# Output Format

## LLM Visibility Optimization Plan for [Product Name]

### 1. Page Structure for Query Fan-Out

- [Your product-specific suggestions]

### 2. Trust Signal Enhancements

- [Your product-specific implementation ideas]

### 3. Crawler Accessibility

- [Your technical SEO suggestions]

### 4. Monitoring & Iteration

- [Your advice for daily tracking and fast iteration]

### 5. Why This Matters

- [Remind the user what’s at stake with AI-driven visibility]

# Final Prompt

Please describe your product in 1–2 sentences. Include what it does, who it serves, and what content types support it (e.g., product pages, blog posts, help docs, videos).

---

Everything is being rebuilt!

We’re in the messy middle of these changes. We’re losing traffic from channels that have been a staple of our online strategies, we’re playing defence to keep as much as we can and it’s not clear how we go on the offence after the new things.

But, there is going to be so much opportunity for the people who figure out that new AI-driven playbooks.

Here’s my honest take, I had reached, what I felt, was the end of my career in growing revenue for tech brands a couple of years ago. I’d done a lot, the playbooks were known and most of your time was optimisation of the things we knew and hiring the best people. I was thinking about my next thing.

AI was that next thing. It’s messy, it’s chaotic, it’s full of unknowns, but it’s empowering. I dedicated myself to knowing everything I could about AI-GTM.

AI has given me the ability to ship and create more than I ever could before.

I launched a product, I have another one coming, launched this Substack, am growing a YouTube channel on AI, creating content on AI, doing AI transformation for HubSpot.

All that to say, AI is going to fundamentally change traditional search, but the opportunity for figuring out new playbooks like AI search optimisation will be huge.

Be curious, ship, create and learn.