Here's a stat that should keep every CMO up at night: 95% of the time, the winning vendor was already on the buyer's Day One shortlist, formed before any seller contact. And the pre-contact favorite wins 80% of deals.
Let that sink in. According to 6sense's 2025 research, four out of five deals are essentially decided before your sales team even knows the buyer exists. If you're still measuring marketing success by MQLs and form fills, you're measuring the 5% of deals that weren't already decided.
The culprit? Or hero, depending on how you look at it: Large Language Models.
The 94% Reality Check
6sense's 2025 Buyer Experience Report, based on nearly 4,000 B2B buyers making purchases with a median value of $300,000 to $400,000, found that 94% of buyers now use LLMs during their buying journey. That's up from 89% the prior year, according to Forrester's Buyers' Journey Survey.
But here's where it gets interesting. The panic narrative says buyers are using ChatGPT to replace vendor engagement entirely. The data says otherwise.
Buyers maintained the same number of interactions with vendors as they did before AI tools became widespread: 16 interactions per person with the winning vendor in 2025, compared to 17 in 2024 and 16 in 2023. LLMs aren't replacing vendor engagement. They're complementing it.
What's actually happening is more nuanced and, frankly, more challenging for marketers to address.
The 60/40 Shift Nobody Predicted
For three years, 6sense tracked what they called the "70/30 journey": buyers completed roughly 70% of their evaluation independently before first contacting a vendor. In 2025, that shifted to a "60/40 journey." Buyers now reach out about 3.5 weeks sooner, at 26.4 weeks into their buying journey versus 30 weeks.
Wait, buyers are reaching out earlier? That seems backwards if LLMs are supposedly making vendor content obsolete.
The explanation is counterintuitive but makes perfect sense once you think about it. LLMs are accelerating the research phase, not extending it. The average B2B buying cycle compressed from 11.3 months in 2024 to 10.1 months in 2025, driven by faster AI-enabled research. Buyers are getting smarter faster, which means they're ready to engage sooner.
The catch? They're arriving with preconceived vendor rankings already formed. And if you're not on that initial shortlist, you're probably not getting on it at all. The average shortlist has shrunk to about 2.5 vendors, down from 3.2 a few years earlier, according to Apollo's 2026 data.
Where LLMs Actually Fit in the Journey
Research from Development Corporate reveals something crucial: LLM use doesn't peak at the start of the buying journey. It peaks in the middle. Buyers aren't primarily asking ChatGPT "who are the top vendors for X software?" as their first move. Instead, they use LLMs after they've already identified a shortlist, to compare offerings side-by-side, synthesize vendor documentation, model costs, draft RFP language, and build implementation plans.
This creates two separate visibility problems. First, being surfaced during initial vendor discovery queries. Second, being accurately represented when a buyer in active evaluation asks an LLM to compare you against competitors.
Semrush's survey of 600+ U.S. B2B professionals found that 92% say AI has shaped their vendor shortlist, with 45% saying it did so significantly. And here's the kicker: brand recognition barely registers. Only 7% say they notice a vendor in an AI response because they recognize the name. What makes a vendor stand out is how precisely it matches the buyer's use case.
The Dark Funnel Just Got Darker
Traditional attribution captures only about 27% of the B2B buyer journey. The other 73% happens in what we call the dark funnel: channels with no trackable signal. LLMs have made that funnel even darker.
Research from Muckrack shows that more than 95% of links cited by AI models come from earned media and third-party sources, not vendor websites. Your carefully optimized product pages? They might be invisible to the AI that's shaping your buyer's shortlist.
Responsive's research found that 32% of buyers now use generative AI as much as or more than traditional search engines when researching vendors. In the tech industry specifically, that number jumps to 80%.
What Actually Wins Deals Now
Here's where it gets practical. Demand Gen Report's research found that industry expertise (52%) now outranks price (49%) and product fit (46%) as the most significant factor in final vendor selection.
Read that again. Expertise beats price.
This makes sense when you consider how LLMs work. They synthesize information from across the web, prioritizing sources that demonstrate depth, credibility, and specificity. Generic marketing copy doesn't surface. Detailed technical content, case studies with real numbers, and thought leadership that actually says something do.
Harvard Business Review's analysis of B2B firms across pharma, precision engineering, and banking found that LLMs are changing go-to-market strategies across industries. The companies winning aren't necessarily the ones with the biggest ad budgets. They're the ones with the most citable, synthesizable content.
The Playbook Shift
So what do you actually do about this?
First, stop gating everything. As one commenter on LinkedIn put it:
AI is forcing us to rethink content gating. If buyers are using AI to research suppliers before engaging, then making expertise accessible may be more valuable than collecting a name and email address. You can't be considered if you can't be found.
Second, think about Answer Engine Optimization, not just SEO. The Fletcher Group's analysis suggests that the old B2B buying process, dominated by Google search and blue link clicks, is giving way to a new process focused on LLMs and AI-enabled search platforms.
Third, invest in earned media and third-party validation. If 95% of AI citations come from sources you don't control, your PR and analyst relations strategy just became a lot more important.
Fourth, prepare your sales team for buyers who arrive later but more informed. They're not exploring options. They're validating a decision they've already made.
Marketing has always been about being in the right place at the right time with the right message. The places have changed. The timing has compressed. And the message needs to be specific enough for an AI to cite it.
The buyers are out there right now, asking ChatGPT about your category. The only question is whether your name comes up, and what it says when it does.
