AI Might Be Recommending Your Competitors. You Can't See It.

By Chad Agozzino, Senior Product Manager, BlueConic

I spend most of my time building products for B2C marketers. Recently, I started having a different kind of conversation with our customers. Not about segmentation or campaign orchestration, but about something most of them hadn't thought much about yet: what happens when a consumer asks ChatGPT, Gemini, or Perplexity to recommend a product in their category.

The short answer, for nearly every brand I talked to: they had no idea.

The traffic is there. The visibility isn't.

Here's what surprised me. The brands I spoke with aren't ignoring AI. They're tracking it. They can see in their analytics that an increasing share of traffic is arriving from AI surfaces. One global athletic brand told me they monitor AI-referred visits the same way they monitor Facebook or email referrals.

But that's where it stops. They see the traffic showing up. They can't tell whether the AI recommended their product, or recommended a competitor and the consumer clicked through to compare. They don't break it out by platform. They don't know which queries are driving it. And they definitely don't know what the LLM is actually saying about their brand when a consumer asks.

The question I kept hearing was some version of: "We can see this channel growing, but we don't know if we should treat this traffic differently. Should we be personalizing for it? Is there information we can pull from the LLM interaction to make the experience better? We just don't know."

That gap between "we can see AI traffic" and "we understand what's happening inside the AI conversation" is where every brand I talked to is stuck right now.

Every brand wanted to know their share of voice. Almost none could measure it.

The single most common question I heard was about visibility. This might sound like "If I sell running shoes and someone asks an LLM for running shoe recommendations, am I showing up? How often? Who's showing up instead of me? Are there categories where I'm winning and categories where I'm invisible?" Sound familiar?

These are reasonable questions. They're also, for most brands, completely unanswered.

The closest tool most people referenced was Semrush, which makes sense because the SEO industry has been tracking search visibility for years. But here's the problem: SEO is a specialized discipline. Many B2C brands don't even have an internal SEO person. They outsource it to an agency. The platforms that track search visibility are built for that specialist persona. They're technical, they're deep, and they assume you already know what you're looking at.

What I kept hearing from brand marketers was something different. They wanted a way to understand, at a high level, how their brand is showing up across AI platforms without needing to become an SEO expert or wait for their agency's quarterly report. They wanted to check their AI visibility the way they check their social metrics: quickly, on their own terms, with enough context to know whether things are getting better or worse.

That tool doesn't really exist yet. The SEO platforms are adding LLM visibility features, but they're building for their existing audience. Brand marketers are being left out of a conversation that directly affects their revenue.

The real problem isn't bad recommendations. It's that brands have no idea what’s being said about them.

When I first started exploring this space, I assumed the issue was that LLMs give bad product recommendations. After running hundreds of queries across multiple platforms, I'll be honest: the recommendations aren't terrible. LLMs are actually pretty good at synthesizing available information and returning reasonable options.

The problem is that brands have zero control over the inputs. The LLM is pulling from whatever it can find: cached product pages, outdated pricing, descriptions written by third-party sites, stock information that's weeks old. The recommendation might look fine to the consumer, but the brand has no idea whether the LLM is showing current pricing, whether it's recommending a product they actually want to push, or whether the description matches what they'd say about their own product.

One conversation that stuck with me: a brand marketer described seeing AI-referred traffic growing in their analytics, but they couldn't answer the most basic question about it. Did the AI recommend us? Or did it recommend someone else, and the consumer came to our site to compare? That distinction matters enormously for how you interpret and act on that traffic. And right now, nobody has the answer.

Two things would fix most of this. Almost no one is doing either.

In every conversation I had, two gaps stood out.

Gap #1:

The first is connecting product feeds. This is the simplest, highest-impact action a brand can take, and almost nobody is doing it. When you connect your product feed to an LLM platform, the AI starts working with your actual product data: real descriptions, current pricing, live inventory, accurate variant information. Without it, the LLM infers your product details from whatever it can scrape. That means the consumer might see last season's pricing, a color you've discontinued, or a description that a reseller wrote.

Connecting a feed isn't a massive technical lift. It takes minutes in some cases. But it requires knowing it's an option, and most brand marketers I talked to didn't. The brands that figure this out early will own the accuracy of their AI recommendations. Everyone else will keep hoping the LLM gets it right on its own.

Gap #1:

The second is building a presence on LLM platforms. No brand I spoke with had built a ChatGPT app. No one had created a Gemini Gem. The appetite wasn't zero, but the readiness was. Brands aren't sure how to build one, aren't sure if they need one, and aren't sure what they'd use it for. That's a knowledge gap, not a lack of interest.

This is where the opportunity gets interesting. For the brands that move first, building on these platforms means owning the conversation. Instead of hoping the LLM recommends you accurately, you create the space where the recommendation happens on your terms. You can see every query, every response, every product suggestion. The black box becomes transparent.

AI discovery is consolidating fast. Early movers will be rewarded.

Every major discovery channel has followed the same pattern. Early movers establish positions that become permanent. Latecomers spend years trying to catch up.

Search followed this pattern over a decade. Social compressed it to about five years. AI discovery is moving faster than either. The infrastructure is forming right now. OpenAI and Stripe launched a protocol for instant checkout inside ChatGPT. Google shipped its own commerce protocol into Search AI Mode and Gemini. The credit card networks created a framework for governing what AI agents can spend and where.

These aren't announcements to watch. They're plumbing being installed. The rails that will carry consumer purchasing through AI platforms are being laid down in real time.

For B2C brands, the math is straightforward. If consumers are increasingly asking AI for product recommendations, and the brands that show up with accurate, current, well-structured product data get recommended more often, then the cost of waiting is measured in lost share of voice that compounds every quarter.

The brands I talked to all recognized this intuitively. None of them had a plan for it yet.

Where AI Discovery is going.

I'm a product person, so I think about this in terms of what we can build and what we can measure. What I took away from these conversations is that the market needs three things, roughly in this order.

1. visibility. Give brand marketers a way to see how they're showing up across AI platforms, who's getting recommended instead, and where the gaps are. Not an SEO tool with LLM tracking bolted on. Something built for the person who runs the brand, not the person who runs the technical SEO audit.

2. explainability. Don't just tell me I'm losing. Tell me why. You're not showing up for this query because your product descriptions are too thin. You're losing to this competitor because they have a more structured FAQ section. Here's what you could change, and here's the first draft to get you started.

3. control. Once you can see the landscape and understand the gaps, give brands the tools to act on it. Connect their product feeds. Structure their data for LLM consumption. Build the presence that turns AI from an uncontrolled channel into one they can optimize.

We're building toward all three at BlueConic. But the first step isn't a product pitch. It's a wake-up call. If you're a brand marketer reading this, go ask ChatGPT to recommend a product in your category. Look at what comes back. Check the pricing. Check the descriptions. Check whether you show up at all.

That ten-minute experiment will tell you more about the state of your AI visibility than any report I could write.