Here’s a scenario worth sitting with.
A senior IT leader at a mid-market company is evaluating cybersecurity vendors. She opens an LLM, describes her problem, and asks for an initial shortlist. Four or five names come back, and then she spends the next hour going deeper in the same conversation: pricing comparisons, customer reviews, how each vendor handles her specific use case, what their customers say about implementation. By the time she closes the chat, she has a pretty clear picture of who she’s going to call.
Your sales team has no idea this is happening, as there’s nothing in the CRM. No intent signal, no form fill, or ad click. But she has already decided whether you’re worth talking to.
This is the new front end of the B2B buying journey, and most marketing teams are not set up for it.
The Buyer’s Journey Starts Before You Know It
At our customer event last month, a session on buying behavior and LLM visibility drew on research conducted across seventy thousand data points, segmented by industry, to understand what’s actually driving brand visibility in AI-generated responses. A few things jumped out.
Eighty-nine percent of B2B buyers are now using AI tools as a primary source of information at some stage of the buying process. More importantly, the data suggests that the initial shortlist, the four or five vendors a buyer starts with, is increasingly being formed in an LLM before the buyer ever visits a website. If you’re not on that shortlist, there’s a reasonable chance you’re out of the conversation before it formally starts.
The flip side, and this is the part I found genuinely interesting, is that buyers who have done this research arrive at your website in a different state. They’re not browsing. They’ve already been through something like a sales cycle. They come ready to buy. The problem is that most websites are built for people who are still deciding whether they have a problem, not people who have already decided they want to solve it.
The AI Shortlist Is the New First Impression
Here’s where it gets more complicated. LLMs are not neutral aggregators. They are skeptical, and they apply that skepticism unevenly.
Your brand narrative, the language you’ve carefully crafted about who you are and what you stand for, is treated essentially as a claim that needs external verification. What you say about yourself matters much less than what others say about you. The research found that third-party content: earned media, analyst coverage, G2 reviews, executive posts on LinkedIn and Reddit, carries significantly more weight than owned content in driving LLM visibility.
The trust hierarchy, roughly, goes: independent verified sources first, then vendor content that is specific and factual about what the product actually does, then broader brand narrative, and then paid content, which LLMs tend to deprioritize or treat with visible skepticism.
What that means practically: you can publish all the thought leadership you want, and it may not move your LLM visibility much. But a strong G2 profile, a piece of earned media in a respected trade publication, or an analyst citing you by name in a report? Those carry weight that no amount of owned content can replicate.
Every Industry Has a Different Visibility Playbook
The industry-specific nuance here is also worth paying attention to.
In cybersecurity, earned media was the dominant driver of LLM visibility in the research we reviewed at the event. In cloud technology, analyst coverage had an outsized impact that it didn’t have in other categories. The implication: there is no universal playbook. The right signal depends on your industry, and the answer is different enough across sectors that a generic “improve your SEO” approach won’t get you there.
The other thing that came up clearly: consistency across channels matters more than most teams realize. LLMs are very good at finding inconsistencies in your brand story. If your LinkedIn says one thing and your Glassdoor reviews say something different, the model sees both. If your website claims a capability that your case studies don’t demonstrate, that gap shows up. Buyers interacting with an LLM aren’t just getting a filtered summary of your best content. They’re getting a synthesis of everything that exists about you, including the things you’ve neglected.
What AI Gets Wrong Can Cost You
There’s also a version of this that most marketers haven’t dealt with yet: hallucinations.
LLMs sometimes get things wrong about specific companies. They describe products that don’t exist, cite capabilities that were retired two years ago, or conflate you with a competitor. If a buyer’s first impression of your company is built on inaccurate information surfaced by an LLM, you may not even get the chance to correct it. The question of what an LLM currently says about your brand, and whether it’s accurate, is worth knowing before your buyers find out for you.
The practical starting point is simple: ask. Open ChatGPT, Claude, or Gemini, describe your category, and see who comes up and how they’re characterized. Ask follow-up questions the way a real buyer would. What you find will tell you more about where you stand in the new buying environment than most of the tools in your stack.
Marketing’s New Front Line
The fundamental shift is this: the shortlist used to form during the sales process. Now it forms before it. Marketing has always been responsible for building the conditions under which a deal is possible. That job just got bigger, and the levers are different than they used to be.
Earned credibility now does more work than produced content. Third-party voices now carry more weight than brand voice. And consistency across every channel, including the ones you stopped actively managing three years ago, now affects outcomes in ways that weren’t true before.
This series has been about the gap between what B2B marketing teams know and what their systems, incentives, and measurement actually reward. The confidence gap in the data. The attribution problem underneath it. The question of who really owns the buyer journey. And now the front end of that journey changing in ways most playbooks haven’t caught up to yet.
All of that is upstream of the pipeline. It’s the work that determines whether a deal starts at all.
What happens after it starts is a different and equally important problem. That’s where I’m headed next.