It’s 2004. Your buddy’s older brother just doubled his business overnight by building a website that floods leads into his lawncare business. Suddenly, everyone’s talking about learning HTML, JavaScript, and this new thing called “SEO.”. Maybe you want to build a website for your footwear retail business, but let’s face it, you barely know what “tag” even means outside of a playground game. Even though you might manage to build something, you have no idea how to get it to show up on Google. Tools exist to make it easier, but they’re clunky and nothing like the “magic” that everyone makes them out to be.
Sound familiar?
Welcome to 2025. Now, your buddy’s brother just vibe-coded a SaaS that just hit $1M ARR. Everyone is talking about AI, MCP, and Agentic commerce. You just want to figure out how to keep customers coming to your site once they start asking ChatGPT where to buy the best Christmas present for their in-laws. There’s a million and one tools to use (like your buddy’s brother’s SaaS), but all of them overpromise and underdeliver.
Is this a tech fad that will pass? Should you even care? Do you have to rebuild everything so Claude can see your products?
Let’s talk about it.
So, How do I Optimize for AI?
Onsite AI Optimization
There are many areas to look at when optimizing your site to show up in LLM searches, but the most important elements to directly optimize on your site are on-page content, structured data, product information, and overall site performance. Equally important are things off the site, such as social media content or brand discussions in public spaces, like Reddit.
First, there are a couple quick wins that can help at least start the process:
Ensure you explicitly allow the top AI bots and crawlers in your site’s robots.txt.
You can find a larger list of bots in a number of places, but one such list can be found on GitHub
Your robots.txt should look something like the following (with whatever bots you want to explicitly allow):
User-agent: OAI-SearchBot
User-agent: ChatGPT Agent
User-agent: ChatGPT-User
User-agent: Claude-User
User-agent: Claude-Web
User-agent: ClaudeBot
User-agent: Devin
Disallow:
- Get yourself on the notification list for ChatGPT’s product feed submission. Once it opens, you’ll be able to give your product feed to ChatGPT, which will ensure anyone searching for your products will have accurate and up-to-date listings.
These actions will ensure you’re prepared for the latest updates while keeping you ahead of the curve with ChatGPT and how it will be used for shopping.
Next, it’s more important now than ever to make sure your site has clean and accurate structured data (see Google’s article explaining how structured data works for SEO for more info). Previously, this was one of the ways Google (or any search engine) understood the content of the webpage and was able to create rich snippets for shopping listings.
With the advent of AI, this data is even more vital. AI crawlers can develop a better understanding of a domain when content is appropriately structured and has complete data. Any AI agent can use the structured data on a site to understand the listings similar to how Google understands the listings.
Finally, what could arguably be the most important is on-page content. People are already using AI as a search assistant, so any marketing content has the potential to be scanned, indexed, and understood by AI. What’s important is to tailor your content towards topics that are relevant to the customer.. If your customers typically search for “best wrestling shoe under $80”, then make sure your marketing content is optimized for that.
The important distinction between this and SEO is that the content is optimized for the longer query, and not just keywords. When using Google or a traditional search engine, someone might search something like “best wrestling shoe under $80”, while someone using an AI assistant might have a much longer and more conversational query:
“I’m a beginner wrestler. I’m looking to purchase some wrestling shoes that are comfortable, have good grip, and are very durable. My budget is $80. I need the shoes ASAP. Please find shoes that I can buy at a store near me today or have less than three days shipping.”
Given something like this is much more likely to represent a search made with an AI agent, writing marketing content that is tailored towards more longtail searches could be much more important to showing up with AI. Articles such as “The Best Wrestling Shoes for Budget-Conscious Shoppers” or “Budget Wrestling Shoes that will Last for Years” would be perfect examples of content tailored to such user queries. Just like a traditional search user, AI will likely read articles and blog posts before it ends up on a PDP.
Some people may still use the same search query for an AI assistant that they would for a traditional search engine (i.e. “best wrestling shoes under $80”). In cases like these, all of the above are still important, as LLMs search for a much wider context than traditional search engines typically do. The LLM might find a few options, but then ask the user follow up questions to help narrow their search and glean additional context.
Offsite AI Optimization & Marketing
Like traditional SEO, there are important things that have to happen off of your site in order for your site to show up in AI searches. When optimizing for SEO, it’s important to have links to your site from external sources. To optimize for AI assistants, it’s important to have your brand simply mentioned by other sources, particularly in a positive light.
Mentions like this could come from a variety of means, from reviews on blogs to social media mentions to Reddit threads. Brand identity and visibility are absolutely vital in this space, as AI assistants will pull from all sorts of public information to determine the best fit for the user’s query. No matter how amazing your website is, if nobody likes your product or brand and they’re often talked about negatively on the internet, it’s unlikely that AI assistants will return your products to shoppers.
Isn’t AI Optimization Just SEO with More Steps?
There is a LOT of crossover between SEO and optimizing for AI, but the context for the optimizations are very different.
SEO thinks about keyword combinations, search relevance, and site quality. AI agents think about how to best solve the user’s problem using these same elements, plus a host of new aspects that are changing almost daily.
On the surface the work might seem similar, but looking at it from this angle, it becomes clear that a much more nuanced approach is needed for AI-optimized content. According to an article by Andrew Holland, traditional SEO:
focuses on product descriptions
optimizes for product features
centers on keyword-based searches
In contrast, an AI-optimized approach:
focuses on problem-solving capabilities
optimizes for Category entry points (more in Andrew’s article about what these are and why they’re important)
centers on user scenarios
focuses on brand positioning
focuses on product and brand attributes
Clearly, a much more holistic approach is needed to ensure a site is properly optimized for AI. A brand needs to have a fully cohesive identity, clearly demonstrate value, and maintain a positive reputation in order to be recommended by LLMs. LLMs look at intent: the why, when, how, and where about the user’s specific situation, feelings, and desires, and then use what they’ve gathered to present a solution.
Final Thoughts on GEO & AI Optimization
AI isn’t going anywhere. It’s no Skynet or Ultron, but it’s here to stay and growing rapidly. Our best course of action is to lean into it and prepare ourselves for the potential of a day where it becomes the standard interface for the internet.
We are still very much in the wild west days of AI. The industry is constantly evolving and shifting. At CQL, we strive to give our clients holistic commerce solutions, including AI readiness. AI isn’t a silver bullet that will destroy the world or do our jobs for us just yet, but it is certainly important for us to stay ahead of the curve.
To learn more, please view our AI Services or contact us.
