The Rise of AI Discovery Engines (and the ChatGPT Effect)
It’s hard to overstate how quickly AI chatbots have entered the mainstream. ChatGPT attracted over a million users in its first week, and usage has only surged since. As of 2025, ChatGPT is reportedly handling over a billion search-like queries per week Consumers are using it to get answers, advice, and yes – product recommendations, faster than they could achieve themselves. OpenAI has even rolled out shopping features that provide personalized product suggestions, price comparisons, reviews, and direct buy links, effectively turning ChatGPT into a full-fledged shopping assistant. This trend, dubbed “conversational commerce,” means people can ask “What’s the best budget-friendly standing desk?” and get a curated answer in-chat, rather than combing through dozens of sites, facets and filters, comparison sites, articles, products and reviews.
Critically, LLM-powered search is becoming a true alternative to traditional search engines. A recent Semrush/Statista report found that 13 million U.S. adults used generative AI as their primary search tool in 2023, and that number is expected to explode to over 90 million by 2027 (martech.org). That’s a massive shift in how consumers find information. It’s no wonder we’re seeing early signs that even Google’s dominance is feeling the heat.
Google’s Search Dominance Is (Finally) Slipping
For the first time in a decade, Google’s global search market share has dipped under the 90% mark. StatCounter data shows Google hovering around 87–89% share in late 2024 – down from the ~90%+ it enjoyed before ChatGPT burst onto the scene. In the U.S., Google’s share fell from about 90.3% in November 2024 to 87.4% by December. A few percentage points might not sound like much, but at Google’s scale this represents billions of searches moving elsewhere. It’s a massive shift in user behavior.
Why the dip? One likely factor is the rise of AI chat tools siphoning certain queries. Microsoft’s Bing (which integrated ChatGPT into its search) has gained a bit of share, and new AI answer engines are emerging. In short, consumers have more choices now for finding answers. Instead of asking Google and sifting through results, they can get a single conversational answer from an AI. This “ask me anything” convenience is starting to pull users away from the Google habit. For e-commerce brands, this signals that visibility can no longer depend solely on Google. If buyers aren’t searching the old way as much, you need to ensure your products are discoverable where they are searching.
Why Traditional SEO Isn’t Enough Anymore
SEO isn’t dead, but it’s no longer sufficient on its own. If shoppers are skipping the search results page entirely and asking ChatGPT or similar AI for what to buy, your stellar Google ranking won’t help you reach them. As one eCommerce SEO expert put it, “shoppers are asking AI tools directly for product recommendations because they’re skipping the search results page entirely. If your brand isn’t showing up inside ChatGPT, you’re already missing sales opportunities you can’t see in your usual SEO reports.” In other words, there’s a blind spot in your marketing if you’re not accounting for AI-driven discovery.
Traditional SEO focuses on getting your site to rank on a search engine results page (SERP). But an LLM like ChatGPT doesn’t “rank” websites in the same way or always show the source unless prompted. Often, it digests information from across the web and delivers a single, synthesized answer. If your product information isn’t part of that synthesis, the AI might recommend a competitor or a random alternative. We’ve already seen instances of ChatGPT recommending niche brands that barely show up in Google, simply because the AI found some data that made those products seem like a good fit. The playing field is changing: generative AI can surface brands and products that traditional search might overlook.
The upshot is that companies must optimize not just for Google’s algorithm, but also for AI algorithms – a practice some are calling “Generative Engine Optimization.” It means thinking about how an AI selects and trusts information. What data does it have about your products? Is that data accurate, up-to-date, and compelling? If you haven’t provided it, the AI might rely on third-party content or outdated info from its training data. That’s a risky scenario. As one marketing strategist warned, “GenAI is already telling your brand’s story — with or without you”, so you’d better guide it by feeding the AI the right story.
Chatbots: The Next Critical Marketing Channel for Product Discovery
Instead of viewing AI chatbots as a threat, savvy eCommerce leaders see an opportunity: LLM-based product recommendations are a powerful new customer acquisition channel. When an AI like ChatGPT suggests “the best noise-cancelling headphones” and mentions your product by name, that’s as good as a word-of-mouth referral – but coming from a trusted digital assistant. Early evidence shows this channel can drive serious results. In fact, brands appearing in AI-powered searches are gaining a massive edge: sales conversions driven by ChatGPT recommendations have skyrocketed by 436%, according to a Digiday analysis. In other words, customers who arrive via an AI recommendation are highly likely to convert, because they come in with a high degree of trust and intent.
This trend isn’t limited to tech early adopters. As millennials and Gen Z shoppers (who are very comfortable with AI tools) gain buying power, the influence of chatbot recommendations will only grow. By 2025, over 65% of decision-makers in the B2B space will be digital natives raised on AI and social searc – and consumer retail is not far behind. These generations are perfectly happy to ask a chatbot for product advice and follow its guidance. For eCommerce marketers, this means chatbots have become “the new SEO” in terms of capturing demand. If you’re not being recommended in that AI-driven conversation, you simply won’t be in the running for the sale.
Another compelling aspect: AI recommendations aren’t pay-to-play (at least not yet). Unlike Google search results or marketplace listings where ads and paid placement dominate, ChatGPT’s shopping feature currently chooses products independently, not based on ad spend. This levels the playing field. A smaller brand with a great product and a strong AI-visible presence can outrank a big brand, because the AI is looking for best answers, not biggest budgets. This is a golden window for up-and-comers to gain visibility. Less-established brands can compete on a more level field in AI-driven discovery, as one MarTech report noted. But that only holds true if you’ve done the work to make your product data AI-accessible and convincing.
Structured Data: Making Your Products Visible and Understandable to AI
So, how do you actually ensure your products show up in AI recommendations? The first step is to help the AI LLM scrapers. Large language models don’t magically know about your latest inventory or the details on your product pages – they rely on the data they’ve been trained on or can actively crawl. This is where structured data comes in. By using formats like JSON-LD schema markup on your product pages, you provide a clear, machine-readable feed of your product information (price, availability, features, reviews, etc.). Think of it as putting up a neon sign for any AI that says, “Here’s exactly what this product is and why it’s relevant.”
Modern AI systems are getting better at leveraging this structured data as a factual backbone. Recent advancements mean LLMs can incorporate and understand schema markup rather than treating it as just random text. In essence, LLMs don’t have to “guess” when you supply them with structured facts – they can directly pull from that JSON-LD knowledge. For example, if your page clearly annotates a product as “Wireless Noise-Cancelling Headphones, $99, 4.5 stars, 2000 reviews,” an AI doesn’t have to infer those facts from prose; it’s served to them on a platter. This boosts your credibility in the AI’s eyes (and algorithm) because the information is unambiguous and authoritative.
Beyond your own website, consider distributing your product data across multiple AI-accessible channels. The more places an LLM can encounter your products, the better. This could mean providing your feed to aggregator sites, participating in marketplaces, or ensuring inclusion in knowledge bases that AI agents draw from. Each of these acts as another “discovery point.” Remember, ChatGPT and similar models were trained on vast swaths of the internet. If your product only exists in one corner of the web, the odds of the AI having seen it (or scoring it as notable) are lower than if your product is mentioned and structured in many relevant places. In short, broaden your footprint: make your product data ubiquitous and structured. As long as those references are legitimate and consistent, you’re training the AI to recognize and trust your brand.
On a practical note, implementing structured data is easier than it sounds. Platforms like Shopify have apps (e.g. Smart SEO or JSON-LD for SEO) that help inject schema markup without you needing to code. It’s worth the effort – you’re essentially future-proofing your SEO for both traditional search engines and the new AI engines. Discovery in ChatGPT isn’t about brand size or ad spend right now; it’s about making your products easy for AI systems to understand, trust, and recommend. Structured data is the technical foundation of that understanding.
OpenAI’s Upcoming Product Feed Integration
Hot on the heels of ChatGPT’s growing role as a discovery engine, OpenAI has just announced (at the time of writing this article in May 2025) that merchants will soon be able to submit product feeds directly. No site crawling or scraping required. Instead of hoping for an LLM to stumble across your catalog, you’ll upload a structured JSON-LD feed of your products straight to OpenAI’s platform. This not only ensures your latest inventory, pricing, and descriptions are accurately represented, but also lightens the load on OpenAI’s end, making the whole process more efficient (and potentially more cost-effective). While the feature isn’t broadly available yet, you can register your interest and share your use case here: https://openai.com/chatgpt/search-product-discovery/. Keep an eye on this integration, it could become the fastest path for your products to surface in AI-driven shopping conversations.
Expanding Your AI Discovery with 37x Marketplaces
So, how can you quickly capitalize on this new paradigm whilst you're waiting for OpenAI to open up their product feed integration?
This is where 37x comes in. 37x specializes in powering marketplaces and product listing hubs that are tailor-made to be crawled by AI and search engines alike. In practice, 37x can help your brand spin up additional online marketplaces (beyond just your website) that showcase your products with complete JSON-LD structured data baked in. These marketplaces are designed from the ground up to be easily navigable by LLMs – meaning an AI agent crawling the web for product info will effortlessly find and digest your offerings.
By deploying many 37x-powered marketplaces, you effectively multiply your discovery points. Imagine having numerous mini storefronts or curated product showcases, all broadcasting rich, structured data about your brand and products. An LLM scanning the web for “best home office desks” might encounter your products on a 37x marketplace that focuses on office furniture, even if it missed your main site. Each such encounter increases the chance that the AI will include your product in its recommendation to users. It’s a bit like SEO in the early days, the more well-placed “signals” you have out there, the more visibility you get, except now those signals are read by AI assistants in addition to human searchers.
Another advantage is consistency and control. 37x ensures that the structured data across all these marketplace instances is accurate and up-to-date, drawn from a single source of truth. You don’t have to worry about outdated info or discrepancies confusing the AI. Every 37x marketplace you power can automatically sync changes in pricing, inventory, descriptions, etc., and present them in clean JSON-LD format. In doing so, you’re not only increasing the quantity of AI-accessible data about your products, but also the quality. The result? When a chatbot scours its knowledge to answer a shopping query, your products stand out as well-documented and credible, available from multiple providers and shopping destinations.
For eCommerce decision-makers, leveraging 37x is a strategic way to accelerate your LLM discovery strategy without reinventing the wheel. You get the benefit of expanded reach (through new marketplaces) and technical optimization (through structured data) in one package. It’s a practical approach to ensure your brand keeps surfacing in the conversations that consumers are having with their AI assistants.
Embrace the Future of Discovery
The era of AI-driven product discovery isn’t some distant future scenario – it’s here now. Consumers are already chatting with bots to find their next purchase, and the numbers show this behavior is growing fast. A small drop in Google’s market share or a niche experiment with ChatGPT today can quickly snowball into a major shift in customer acquisition tomorrow. Forward-thinking eCommerce leaders are preparing now. They’re complementing traditional SEO with AI-focused strategies: integrating with chatbots, structuring their data, and expanding their online presence in ways that AI can pick up.
Adopting an LLM discovery strategy is about staying visible and competitive in the channels where customers are headed. It’s about not letting your brand become invisible as the discovery journey fragments across search engines, voice assistants, and AI chats. The good news is that, with the right approach, you can turn this disruption into an advantage. Brands that move early to optimize for AI-driven discovery can leapfrog competitors, win over the new generation of shoppers, and capture sales that others are missing.
In summary, don’t rely on Google alone to deliver your next customer. Ensure your products can be found and recommended by the latest and greatest LLM chatbots. Leverage structured data and proliferate your product listings through LLM-friendly channels to cast a wide net. And consider partners like 37x to turbocharge this effort by creating AI-crawlable marketplaces for you. By doing so, you’ll position your eCommerce business to thrive in an AI-first discovery era – where your next big source of customers might just be a chatbot’s suggestion away.
Now is the time to act, experiment, and invest in your AI discovery strategy. The consumer’s mindshare is up for grabs in this new channel. Make sure your brand is part of the conversation when an AI is asked, “What should I buy next?”
Frequently Asked Questions
Q1: How do I get my products recommended in ChatGPT conversations?
A: Ensure your product data is easily accessible to AI systems by adding JSON-LD schema markup to your product pages, participating in AI-friendly marketplaces (like those powered by 37x), and distributing your product feeds across relevant aggregator sites. The more structured, up-to-date data points an LLM can crawl, the more likely it is to surface your products in its recommendations.
Q2: How can I let ChatGPT know about the products that I sell?
A: ChatGPT and other LLMs “learn” about new products by crawling the web or ingesting structured datasets. To introduce your products:
- Embed Product schema (via JSON-LD) on your site.
- List your catalog on marketplaces and platforms that the AI indexes.
- Keep your data feeds (price, availability, specs, reviews) synced and accurate so the LLM pulls the latest info.
Q3: Do I still need traditional SEO if I’m focusing on LLM discovery?
A: Yes. Traditional SEO remains vital for web traffic and brand visibility on search engines. An LLM discovery strategy should complement, not replace, your SEO efforts. Structured data and AI-friendly distribution broaden your reach into chat-based channels while SEO continues to capture organic search traffic.
Q4: What type of structured data should I use to optimize for AI recommendations?
A: Use JSON-LD formatted according to the schema.org/Product specification. Key properties include name, image, description, sku, offers (price & availability), and aggregateRating (reviews). This machine-readable format helps LLMs accurately parse and trust your product details.
Q5: How can 37x help with my LLM recommendation strategy?
A: 37x powers specialized marketplaces that automatically generate and maintain JSON-LD structured data for your products. By creating multiple AI-crawlable storefronts, 37x multiplies the entry points where LLMs can discover and recommend your products—accelerating your visibility in chatbot-driven shopping journeys.