Consumers are changing the way they discover products online, not through flashy ad channels or organic rankings alone, but through the quiet authority of large language models (LLMs). This is why everyone is talking about generative engine optimization, or AI search.

When a user asks ChatGPT about the best protein powder for sensitive stomachs or for help finding a sneaker suitable for long-distance trail running, they’re not browsing 10 links. They’re seeking answers. And ChatGPT delivers, with summaries, product picks, reviews condensed into insights, and buy buttons inside the result.

You read that right: buy buttons inside the result. 

Image courtesy of OpenAI

ChatGPT’s answer is the new first impression. And increasingly, it’s also the last.

If you're a marketer like me, there's a good chance your Slack and Linkedin feeds are buzzing with AI news, strategies, and maybe even fear. Now's not the time to panic. But it's not the time to dig your heels into old marketing methods, either.

It's time to evolve.

This means teaching AI platforms why your product should be recommended and when. 

It means optimizing for both humans AND machines.

It means thinking of your marketing strategy as a layered stack of both old and new.

OpenAI is exploring ways merchants can submit product feeds directly to ChatGPT. In the meantime, take a look at how the marketing playbook is evolving; not as a binary old vs. new, but as a layered stack. The brands that thrive today (and tomorrow) will be the ones that master both.

Give LLMs all the context they need

The core job of marketing has always been to show up at the right time, in the right place, with the right message.

With LLMs, this requires one extra leap: making sure the AI understands what problem your product solves, and when it’s the best choice to solve it.

LLMs don't index based on keywords alone, they infer context. That means your product doesn't just need to say what it is. It needs to say why it's relevant, for whom, and in what scenario.

This is where product-market fit becomes more than a business principle—it becomes a language challenge.

Instead of just writing “wireless earbuds,” you write:

“Designed for remote workers who need all-day comfort and crystal-clear mic quality. Ideal for Zoom meetings and noisy cafés.”

LLMs use that context to understand: this product is not for runners or audiophiles, but for professionals working remotely. That’s when it gets selected in AI-driven recommendations.

To support this:

• Use structured schema (Product, Review, FAQ) to expose product attributes cleanly

• Create content that mirrors real user intent: who it’s for, when to use it, what alternatives it replaces

• Fill product feeds with detailed descriptors, not just marketing copy

This isn’t SEO. It’s AI comprehension.

Think entity graphs over page rankings

LLMs don’t “rank websites.” They build networks of meaning.

Think of them as a giant web of entities: products, brands, use cases, ingredients, certifications, reviews, and problems. These aren’t ranked by keywords; they’re linked by relevance, authority, and relationships.

Your product isn’t just a page. It’s a node in a knowledge graph.

To plug into this graph:

• Use consistent product names and attributes. LLMs recognize consistent copy the same way humans recognize consistent visual branding. And don’t be too cutesy. Define what you do best, then articulate it the same way across channels (on-site, Amazon, TikTok, and so on).

  
  

• Feed the AI structured data about your brand and offerings (schema, reviews, FAQs) so your messaging is super LLM-friendly.

  
  

• Link products to broader concepts (e.g. "sustainable," "made in the USA," "ideal for gifting"). This makes it easier for AI to understand when and why your brand should surface, especially when users ask open-ended questions, like “what’s a thoughtful gift under $50 for a coworker? Please, no mugs.”

Test how your products show up across platforms

Google SEO is still absolutely essential. But it’s no longer the only place you need to rank.

Now, you want your product to surface:

• In a ChatGPT Shopping recommendation card

• As one of three product picks in a Google AIO result

• In a Perplexity shopping sidebar

• Via Bing Chat’s affiliate summaries

• In a Reddit thread the AI summarized for someone

  
  

Each of these is a new surface—and each has its own rules Yes, it can feel overwhelming, but this is a good place to start:

• Maintain traditional SEO hygiene (titles, meta, speed, backlinks)

  
  

• But also ensure you’re discoverable in LLM-native surfaces: submit product feeds, use structured data, allow AI crawlers, and test how your products show up in AI answers.

Visibility now requires multi-surface optimization: not just ranking, but being selected across every interface.

Go beyond affiliate clicks and create AI learning loops

Here’s something not enough marketers realize: LLMs learn from affiliate and influencer content. That blog post comparing “top travel backpacks” might be scraped, summarized, and echoed back by an AI assistant. The YouTube video breaking down “budget noise-canceling headphones” becomes part of the model’s context when someone asks, “what’s a good pick under $100?”

This makes affiliate and influencer content not just performance drivers, but also training data.

This is a very powerful leveler for emerging brands.

If you don’t have thousands of reviews, Reddit threads, or organic chatter about your brand yet, you can buy your way into the conversation. This is not just for clicks, although that will help,  but to teach the AI that your product exists, and why it’s worth recommending.

Start with paid affiliate and creator partnerships that create structured, searchable, and relevant content:

• “Best for X” guides with clear comparisons

• Reviews that use descriptive product names, specs, and use cases

• Content on high-authority platforms (blogs, YouTube, forums)

You ideally want both organic UGC and creator coverage, but if you can’t have both, start where you can control the message with affiliates; LLMs won’t know who you paid to talk about your product and who did so organically.

Lean into the new “media buy”

Ad spend used to be the lever. Now, data quality is.

AI platforms are experimenting with product listings, direct shopping integrations, and affiliate revenue models. That means you don’t just want to show up, you want to show up accurately and attractively.

What to do:

• Submit up-to-date product feeds (Google Merchant Center, Microsoft, and soon, ChatGPT)

• Include complete info: price, inventory, specs, GTINs, high-res images

• Monitor traffic from utm_source=chatgpt.com and similar referrers to measure AI lift

  
  

This is the new “media buy.” You’re not placing ads. You’re placing data—and hoping the AI does the rest.

Marketing has always been about visibility, relevance and timing.

Now you have a new gatekeeper: AI.

It won’t replace your funnel, but it will reshape the top of it. Your job as a marketer is no longer just to earn attention from people; it’s to earn trust from the machines that guide them.

And that starts with rethinking how you speak to them.