To show up in AI search results, you’ll need to make a few tweaks to your SEO program. The good news is if you’re already doing SEO at a high level (like my clients), then you don’t need to change too much.
Here’s how you can optimize your SEO program for LLM-powered AI search engines like ChatGPT, Perplexity, Claude, and the rest.
Before we talk tactics, here’s a quick, expert-level summary for executives:
- Despite the hype, AI search traffic is not a huge driver of website visits. Referral traffic from LLMs is around 1% of total traffic for most B2B SaaS websites. This suggests that while AI is growing, it’s not a primary traffic source you’re missing out on.
- Customers are searching more than ever before. Normal search engine usage is not shrinking, it’s increasing in addition to AI search growing (source).
- If you’ve got a great SEO program today, you’re in great shape for AI (just need a few tweaks). Corollary: If you’re doing basic SEO “best practices”, you’re going to get crushed.
- The principle of SEO is the same: Wherever customers search for information about products/services, we can influence how our brands appear in the results (normal search engines, AI search, etc).
- AI search learns about brands the same way search engines do: They analyze public text and data on the web: Company websites, articles, forums, news sources, and social media platforms. Often, they do a web search just like you.
- We can influence AI search results by creating and promoting consistent messaging (content) about your brand.
- Like all marketing work, we can use AI tools to automate (mainly content generation), but the biggest challenge will always be differentiation.
OK, let’s get into SEO tactics and how they can be updated for AI search.
Content Strategy
Legacy SEO: Keyword research to understand your customers’ needs and publish content which answers their questions.
New AI Search Approach: Extremely top-funnel content is effectively “dead” as LLMs scrape and summarize informational “what is” type content, serving it directly to users without providing citations or clicks. So the focus shifts away from top-funnel informational content, more toward long-tail and lower-funnel content. Keyword research is still valuable to measure customer demand for information but we want to focus more on solving problems that move customers down funnel.
For AI search, another new risk is LLMs will fill in gaps to provide answers. For example, let’s say a customer is trying to find pricing for your B2B SaaS product, and you don’t publish it because your Sales team doesn’t want to. AI search engines will use whatever source they can find (your competitors, forums, low-quality content mills, etc) and might share the wrong price with the customer.
AI search platforms also use something called “grounding,” which means they search the web for reliable sources. So if the answer is already included in the training data, it will respond directly without grounding (no citation or click).
That means your content strategy should based on tareting prompts that recommend products and solutions. We can then use these insights to “ground” our content, making it easier for AI models to find and suggest our solutions.
So wheras with Legacy SEO you had to worry about making sure the top-ranked results were flattering (or at least accurate) to your brand, now you also have to worry about updating your product information across the web. So your content strategy includes legacy SEO tactics, and it just became a lot broader and includes tactics like partnership marketing and outreach. Which brings us to backlinks.
Link Building
Legacy SEO: Backlinks are very important third-party trust indicators. Digital PR, branded mentions, and nofollow links considered secondary “nice-to-have.”
New AI Search Approach: Link building just became even more valuable. LLMs use crawled web data to “learn” about your brand, but digital PR, branded mentions, and nofollow links just gained significance because LLMs assess brand visibility through mentions. If an LLM is trained on seeing your brand across the web, you have a better shot at getting included in LLM-generated responses. The same outreach and placement strategies you were using before are just as valid here.
Data Feeds
Legacy SEO: Structured data (schema markup) has been crucial for years, helping search engines clearly understand and categorize your content. It allows search engines to accurately present rich snippets, product details, prices, reviews, and availability directly in search results.
New AI Search Approach: Structured data remains valuable, but new experimental methods are emerging to better feed data directly to AI search engines. One notable format is MCP (Model Context Protocol) servers, which allow brands to provide real-time structured context to AI models. Currently experimental, MCP servers could become essential, especially for ecommerce brands and businesses managing large inventories or extensive service offerings. By directly serving AI-friendly context, businesses can improve how accurately AI search engines index their products and services, enhancing visibility and driving conversions.
User-Generated Content (UGC)
Legacy SEO: UGC leveraged for “parasite SEO,” tapping into domain authority of platforms like Reddit to rank indirectly and gain clicks. Reviews primarily influential via structured data snippets (rich snippets, star ratings) to boost CTR in SERPs.
New AI Search Approach: UGC now directly educates LLMs about your brand. This is why reputation management and sentiment matters. You need to own all questions around your branded search prompts like reviews, help docs, product comparisons, pricing, etc. This includes 3rd party ecosystem management. Encourage high-quality, informative discussions and reviews on platforms LLMs frequently scrape, including forums, social media, and review sites. Reviews, user feedback, influencer campaigns, and other community marketing programs shape LLM perceptions more directly than legacy search engines.
Keyword Optimization
Legacy SEO: Focus on developing content to exactly match search intent with high-value target keywords and variants.
New AI Search Approach: You still need to create content about your primary subjects. You should already be optimizing that content for complete, clearly articulated answers to common user questions. But LLM-driven AI may make better use of thorough, conversational content as opposed to direct, educational content.
Content Structure
Legacy SEO: Your content needs to look like it was produced by a top-tier publisher. Emphasize clear content structure (headings, subheadings, formatting) for user experience. Bullet points, concise summaries, and definitive statements help users accurately parse and reproduce information in AI-generated answers.
New AI Search Approach: No difference.
Content Freshness and Frequency
Legacy SEO: Fresh content valued mostly for breaking news, trending topics, or competitive keyword rankings. Regular updates beneficial but not always mandatory. Publishing often was required to build up a large, authoritative content library.
New AI Search Approach: Unclear. LLMs don’t currently train very often (ChatGPT’s newest model’s training data is a year old). But there is a good reason to publish at scale because AI search engines will eventually find it. LLMs also integrate up-to-date search results into responses, so regular content updates should keep brands relevant in ongoing conversations.
Local SEO
Legacy SEO: Ranking in the “map pack” or Google Business Profile results is key, but for some queries the standard organic results drive the majority of conversions. Success relies heavily on reputability, local backlinks, and customer reviews. Ensure that your business details are accurate and consistent across local directories, review sites, and social platforms.
New AI Search Approach: LLMs also surface location-specific information. There’s a “map pack” on ChatGPT as of a couple weeks ago. The approach here is very similar, with a slight wrinkle that LLMs pull from sources like Yelp, Google Reviews, and Reddit when forming responses about “best X near me.” So optimizing for traditional local SEO now doubles as optimization for LLM-driven local queries.
Metrics and Reporting
Legacy SEO: Success was measured in traffic, rankings, and click-through rate. Dashboards tracked organic sessions and position shifts. More clicks meant better performance.
New AI Search Approach: Legacy metrics still important, Visibility is the new metric to add to the mix. The goal is brand inclusion in AI-generated answers, not just blue link clicks. Mentions across the web matter more than traffic from any one source. Traditional tools like Google Search Console and analytics platforms don’t show how often LLMs surface your brand. New measurement tools/features are coming out on a weekly basis, but I expect Ahrefs/SEMrush to set the bar here. You probably don’t need new software.
The future of AI search
If I had OpenAI’s first mover advantage, I would be trying to eat Google’s lunch as fast as possible. I’m not sure how quickly they might roll ads into ChatGPT, or whether they are working on more SERP-like experiences similar to Google’s excellent ecommerce product Google Shopping. To compete with Google in search they might have to become more Google-like, further validating legacy SEO approaches.
I run a boutique SEO consulting business in San Francisco, CA. I like to play golf, write a little bit, and argue with my friends.