How GenAI and LLMs are Changing B2B Buyer Research, and How to Respond

Ever since ChatGPT entered the business mainstream, B2B marketing circles have been gripped by low-level panic. “All search traffic is going to disappear.” “Buyers won’t visit websites anymore.” “Intent data is dead.”

Here’s what 6sense’s actual research on nearly 4,000 B2B buyers tells us: Yes, things are changing. No, it’s not the apocalypse everyone fears.

Our 2025 Buyer Experience Report represents three years of longitudinal data tracking how B2B buyers actually behave — not how pundits think they behave. These are real buyers making real purchasing decisions with a median value of $300,000 to $400,000. They’re navigating complex buying committees averaging 10 people, evaluating multiple vendors over nearly a year, and yes, using Large Language Models (LLMs) like ChatGPT and Claude along the way.

But here’s the critical finding that should reshape how you think about AI’s impact: 94% of buyers now use LLMs during their buying journey, yet they maintain the same number of interactions with vendors as they did before AI tools became widespread. That’s right: 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.

This whitepaper cuts through the hysteria with data-driven insights about what’s actually changing in B2B buyer behavior, what’s staying remarkably consistent, and how revenue leaders should respond.

We’ll explore how LLMs work, where they fit in the buyer journey, and why platforms like 6sense, built specifically to capture the “dark funnel” of anonymous buyer research, become even more valuable as AI tools expand.

If you’re a CMO, VP of Sales, or demand generation leader trying to separate signal from noise in the AI era, this report provides the roadmap you need.

Chapter 1: What 6sense data reveals: the 60/40 shift

Let’s start with the headline finding from our 2025 research: Buyers are contacting vendors earlier than before. But not for the reasons you might think.

The numbers tell a surprising story

For three years, 6sense has tracked a consistent pattern we call the “70/30 journey.” Buyers completed roughly 70% of their evaluation independently before first contacting a vendor, then used the final 30% for validation and final selection.

In 2025, that shifted to what we’re calling the “60/40 journey.”

Point of First Contact (POFC) moved from 69% to 61% of journey completion. In practical terms, buyers now reach out to vendors about 3.5 weeks sooner than they did in previous years, at 26.4 weeks into their buying journey versus 30 weeks.

This sounds like what you’d expect if LLMs were making vendor content obsolete, right? Buyers getting all their answers from ChatGPT and only contacting vendors at the last possible moment?

Wrong.

Why buyers are actually engaging earlier

When we dug into the data, we discovered the shift has almost nothing to do with LLM efficiency replacing vendor interactions.

Instead, nearly 58% of buyers told us they engaged vendors earlier specifically to validate how AI capabilities were being implemented in the solutions they were evaluating.

Think about it from the buyer’s perspective. Your company is about to spend $300,000+ on a software solution. The vendor’s marketing materials claim their product now has “AI-powered” features. But what does that actually mean? How was it trained? What data does it use? How much will it cost? What are the privacy implications?

These questions can’t be answered by ChatGPT. They require direct vendor engagement.

As Kerry Cunningham, 6sense’s Chief Research Officer, explains in his analysis “The LLM-Ate-the-Dark-Funnel Hysteria: Let’s All Take a Breath Shall We“:

“For companies experiencing a decline in web traffic attributable to LLMs, it is likely the traffic that has gone missing is not from in-market buyers, but from the professionally curious and from future buyers. This is not to say that marketers should be unconcerned about the decline of web traffic. However, that decline is unlikely to impact current pipeline production.”

What hasn’t changed (and what that tells us)

Here’s what remained remarkably stable in our 2025 data:

• Average buying group size: 10.1 people (compared to 10.4 in prior years)
• Interactions with winning vendor: 16 per person (compared to 17 in 2024, 16 in 2023)
• Vendors evaluated: 5.1 (up slightly from 4.5, indicating more thorough research)
• Vendors with prior experience: 3.9 (meaning 76% of shortlisted vendors are already known)
• Buyer-initiated contact: 79% (buyers still control the timing)
• Winner contacted first: 77% (down from 83%, but still dominant)

These patterns reveal something crucial: LLMs are tools buyers use within an existing evaluation process, not replacements for that process. The fundamental dynamics of B2B buying (committee decision-making, extensive vendor validation, and risk mitigation) remain intact.

The global perspective

Our 2025 research covered buyers across North America, APAC, Europe, and the UK. While regional variations exist (UK buyers contact vendors earliest at 51.5% of journey completion, North American buyers latest at 64.3%), the overall pattern holds globally. B2B buying remains a high-touch, relationship-driven process that requires human validation of high-stakes decisions.

The implications for B2B revenue teams are clear: Don’t restructure your entire go-to-market strategy around the assumption that buyers will stop engaging with vendors. Instead, understand where LLMs fit in the journey and optimize the reality of how they’re actually being used.

Chapter 2: Where LLMs fit in the buyer journey

Understanding when and how buyers use AI tools is critical to adapting your strategy effectively. Our 2025 research provides specific insights into LLM usage patterns that challenge common assumptions.

LLMs complement, they don’t replace

The 2025 Buyer Experience Report found that 94% of B2B buyers use LLMs during their buying process. That’s a staggering adoption rate for any technology, let alone one that’s been mainstream for less than two years. Research from Forrester confirms this trend, showing that adoption went from essentially zero in January 2024 to 89% by mid-2024, with 87% of users reporting better business outcomes from using generative AI.

But here’s what matters more: Despite near-universal LLM adoption, buyers reported the same number of interactions with vendors as in prior years. An average of 16 touchpoints per person with the winning vendor, which is unchanged from 2023 and down just one from 2024’s 17.

Why? Because buyers don’t trust LLMs alone for decisions of this magnitude.

When  buyers use LLMs

Our research reveals that buyers use LLMs primarily in the middle of their journey, not at the beginning or end. This surprised us until we considered buyer experience levels.

In the 2025 data, 85% of buyers had direct prior experience with the vendors they evaluated — the highest level we’ve recorded. For established product categories, buyers already know who the major players are before they start researching. They don’t need ChatGPT to tell them “the leading CRM platforms are Salesforce, HubSpot, and Microsoft Dynamics.” They already know that.

Where LLMs prove valuable is in the comparison and synthesis phase. Buyers ask things like:

• “What are the key differences between Salesforce and HubSpot for mid-market B2B companies?”
• “Create a comparison table of marketing automation platforms by feature set and pricing”
• “What are the implementation timelines for enterprise CRM systems?”

These queries help buyers become more informed before engaging vendors, leading to higher-quality conversations. As HubSpot’s 2024 survey found, 48% of buyers now use AI tools to research software, and 98% of those users say it’s been impactful to their process.

The validation imperative

Here’s the crucial insight: Buyers use LLMs to inform their evaluation, not to make their decision.

The average purchase in our study was $300,000 to $400,000. That’s a significant investment for any organization, and a significant risk for individuals making vendor selections. Even if LLMs were perfectly trustworthy sources of information (which they’re not, as we’ll explore), buyers would still verify everything through direct vendor engagement.

This validation behavior shows up clearly in broader market research. TrustRadius found that 90% of buyers click through to verify information provided by AI tools. They’re using AI to become better-informed buyers who ask more sophisticated questions during vendor interactions, but they’re not putting blind trust into what AI tells them.

Generational adoption patterns

The data does show significant generational differences in AI adoption. TrustRadius’ 2025 research found that Gen Z buyers show 2x adoption rates compared to overall averages, using AI “a lot” at 15% versus 8% across all demographics. With 64% of B2B decision-makers making purchases now Millennials or Gen Z, this behavioral shift will only intensify.

But, even amongst the highest AI adopters, the validation pattern holds. G2’s data shows that AI Power-Users spend 30% of their time on research versus 36% for others — they’re more efficient, but they’re not skipping vendor engagement. They’re compressing timelines while maintaining the same validation touchpoints.

The takeaway for revenue teams: LLMs are accelerating buyer education, not eliminating the need for vendor relationships. The buyers engaging with you are better informed and further along in their evaluation, which should improve conversation quality and deal velocity.

Chapter 3: The trust landscape: Peers over vendors and analysts

Understanding who buyers trust (and don’t trust) is essential to adapting your content and engagement strategy for the AI era. Our research reveals a dramatic trust shift that’s reshaping where buyers look for information.

The decline of traditional authority

For years, B2B buyers relied on analyst firms like Gartner and Forrester to validate vendor selection. That’s changing rapidly.

TrustRadius documented a stunning 60% decline in analyst report consultation since 2022. Only 14% of buyers now consult analyst reports during their research, as buyers increasingly question “pay-to-play” research models where vendors fund the analysts evaluating them.

Vendor content faces similar skepticism. TrustRadius reports that while 87% of vendors believe they are transparent about their products’ limitations, only 37% of buyers agree.

Our 2025 Buyer Experience Report provides context for this skepticism: Buyers consumed an average of 13 content pieces during their journey — eight from vendors, but critically, five from third parties. Those external sources often carried greater weight in final decisions, even though buyers consumed more vendor content by volume.

The rise of peer validation

As trust in traditional authorities erodes, buyers are turning to peer experiences and independent review platforms.

G2’s annual research shows that public product review websites jumped from 13% of buyers’ most consulted source in 2021 to 31% in 2024.

Review sites now outrank analyst firms at every stage of the buying journey. According to G2’s 2024 Buyer Behavior Report, buyers trust independent review sites more than analyst firms during discovery (51% vs. 45%), research (48% vs. 46%), evaluation (50% vs. 46%), and final decision (47% vs. 45%).

This shift connects directly to the 6sense data showing that buyers evaluate an average of 5.1 vendors but already have prior experience with 3.9 of them. When entering a new buying cycle, buyers start with a consideration set built from their experience, peer recommendations, and review site research — not vendor marketing or analyst reports.

How LLMs amplify the trust shift

Here’s where LLM usage intersects with trust patterns: AI tools cite review platforms as authoritative sources.

When buyers ask ChatGPT or Claude for software recommendations, the models often reference G2, Capterra, and TrustRadius as “crowd-sourced intelligence” — treating aggregated peer reviews as more trustworthy than vendor marketing claims. This creates a reinforcing loop where review site presence drives both direct buyer research and AI-mediated discovery.

The dual benefit shows up in recent data: 72% of buyers now encounter Google AI Overviews during their research, according to TrustRadius’s 2025 study, and 40% say AI makes finding information easier, doubling from 2024.

What this means for your strategy

The 6sense research showing 77% of buyers contact the eventual winner first (down from 83% but still very high) reveals the stakes: By the time buyers reach out to vendors, they’ve already formed strong preferences based on peer validation, review site research, and prior experience.

Your vendor content still matters — buyers consume it heavily — but third-party validation increasingly determines whether you make the shortlist in the first place. As we’ll explore in the practical strategies section, building authentic review presence and peer advocacy has become mandatory infrastructure, not nice-to-have marketing.

The trust landscape has fundamentally shifted. Adapt or become invisible.

Chapter 4: The Dark Funnel expands: 3 layers of invisibility

At 6sense, we’ve been tracking what we call the “Dark Funnel” since our founding. It’s the vast majority of B2B buyer research that happens anonymously and outside of trackable channels. Our data consistently shows that only 3% of website visitors self-identify through form fills, meaning 97% of your buyer research is anonymous.

LLMs aren’t creating the dark funnel. They’re adding a new layer to research opacity that already existed and making account-based intelligence even more critical.

Layer 1: Anonymous open web research

The first layer of the dark funnel consists of buyers visiting your website, reading your blog, consuming your content, checking your pricing page — all without identifying themselves. They’re researching, evaluating, and forming opinions. You have no idea who they are, what company they’re from, or where they are in their buying journey.

This layer has been the reality of B2B marketing for years. The only visibility comes from intent data platforms (like 6sense) that can identify companies visiting your site through IP tracking and account matching.

Our 2025 Buyer Experience Report confirms this dynamic remains unchanged: Buyers conduct extensive independent research before engaging vendors, consuming an average of 13 content pieces (8 from vendors, 5 from third parties) across their journey. The vast majority of these interactions happen anonymously.

Layer 2: Closed networks and private channels

The second layer is even darker: research happening in channels where tracking is fundamentally impossible.

This includes:

• Private Slack workspaces where buyers discuss vendors with peers
• Microsoft Teams channels for internal buying committee deliberations
• Reddit direct messages requesting product experiences
• LinkedIn private conversations with people who’ve used the product
• Phone calls with trusted colleagues at other companies
• Professional community discussions (Pavilion, Revenue Collective, CMO Coffee Talk)

Research shows that 84% of content sharing occurs through private channels like these — what we’ll call “dark social.”

Our 2025 data shows the scale of this challenge: With average buying committees of 10.1 people, each conducting research independently across multiple channels, you’re looking at 100+ research interactions per deal — the vast majority outside the visibility of traditional analytics and attribution tools.

Layer 3: LLM-mediated research

Now add a third layer: conversational research with AI tools that creates zero engagement signals.

When a buyer asks ChatGPT “What are the best marketing automation platforms for mid-market B2B SaaS companies?” and receives a synthesized answer, no click occurs. No website visit. No form fill. No intent data capture. No retargeting pixel fires. Zero signal reaches any vendor mentioned in the response.

The buyer has progressed from awareness to consideration without creating a single trackable datapoint.

This “automated research assistant” dynamic pushes first sales engagement later while simultaneously reducing the visibility that made traditional demand generation measurable. Even Google search — historically at least trackable through keyword and CTR data — now ends without a click 58.5% of the time, according to Search Engine Land’s June 2025 analysis.

Buying committee complexity

The dark funnel challenge multiplies when you consider committee dynamics. Our 2025 data shows:

• 10.1 average buying group size
• 16 interactions per person with the winning vendor
• 5.1 vendors evaluated

That’s 160+ interactions per deal with your company alone if you win, many multiples of that across all evaluated vendors, and thousands more across independent research channels.

Each committee member might:

• Use different LLMs (ChatGPT, Claude, Perplexity, Gemini)
• Ask different questions based on their role and concerns
• Conduct independent research through different channels
• Share insights in private channels you can’t see
• Form opinions that influence the group consensus invisibly

From a vendor perspective, you’re trying to influence a distributed research process happening across dozens of people, hundreds of touchpoints, and multiple platforms — most of which are completely opaque to traditional marketing measurement.

Why this matters for strategy

The expanding dark funnel has critical implications:

1. Last-click attribution is dead. When 97% of research is anonymous and the buying committee conducts 4,000+ interactions across dark channels, measuring which touchpoint “caused” a conversion is meaningless fiction.
2. Individual lead tracking misses the reality. B2B buying happens at the account level with committee decision-making. A single form fill represents one person from a buying group of 10, and probably not the most influential one.
3. Account-based intelligence becomes mandatory. You need systems that aggregate signals across the entire buying committee, identify accounts showing buying intent before explicit engagement, and predict propensity to purchase based on behavior patterns across anonymous visitors.
4. Predictive analytics compensate for signal loss. As individual touchpoints become harder to track, models that identify high-propensity accounts based on aggregated signals become more valuable, not less.

This is exactly what 6sense was built to solve, and why platforms designed for individual lead tracking and last-click attribution struggle as the dark funnel expands. The future belongs to account-based revenue platforms that can pierce the veil of anonymous research and identify buying intent before explicit engagement.

Chapter 5: How LLMs work: Training data and real-time search

To understand why buyers still need vendor validation despite AI tools, you need to understand how LLMs actually function, and their inherent limitations for high-stakes B2B decisions.

The hybrid approach: Historical knowledge and live search

Modern LLMs like ChatGPT and Claude operate through a hybrid approach that combines historical training data with real-time web search capabilities.

Base Model Training provides the foundation. GPT-4 was trained on approximately 13 trillion tokens from sources including:

• Common Crawl (massive web scraping)
• Books and academic papers
• Reddit discussions
• Code repositories
• Wikipedia
• YouTube transcripts

This training data has cutoff dates — the point beyond which the model has no inherent knowledge. GPT-4’s base knowledge cutoff is April 2023, while GPT-4o extends to October 2023. Claude Sonnet 4 (as of October 2025) has the most recent cutoff at March 2025.

Real-Time Web Search fills the gaps. Through a technique called RAG (Retrieval-Augmented Generation), modern LLMs can search the live web, retrieve relevant information, and synthesize it into responses. When a user asks for “recent” or “current” information, the model triggers web search rather than relying solely on base training.

This hybrid creates better currency for current events and recent product launches, but it doesn’t eliminate bias or ensure accuracy.

What LLMs cite (and why it matters)

Understanding citation patterns helps explain why review sites and peer validation have become so critical. Research from Backlinko analyzing 10,000+ queries found that 90% of ChatGPT citations come from search results ranking in positions 21 or higher — not the top 10 traditional SEO positions.

LLMs mine deep into search results for comprehensive answers, favoring specific content types:

According to analysis from Profound, ChatGPT overwhelmingly cites Wikipedia, followed by Reddit, Forbes, G2, TechRadar and Nerd Wallet; Google’s AI favors Reddit, YouTube, Quora, LinkedIn, and Gartner; Perplexity leans on Reddit, YouTube, Gartner, Yelp, and LinkedIn.

What do these platforms have in common? They provide a wealth of structured, comparative data, and they are mostly community-driven.

The implication: Being in traditional top-10 search results isn’t enough for LLM visibility. You need presence across the platforms LLMs trust as authoritative sources — especially Wikipedia, Reddit, YouTube, and review sites. For SaaS companies, focus on sites like G2, TrustRadius, and SourceForge that have active review communities and strong domain authority.

Systematic biases in recommendations

Here’s where LLM limitations become critical for B2B buyers: The models carry systematic biases that can skew recommendations.

Source Bias favors structured data over narrative content. Research from Princeton’s GEO study found that adding statistics improved visibility by 27-36%, citing sources increased it by 20-34%, and including expert quotations boosted it by 20-35%. Conversely, keyword stuffing showed negative results, actually decreasing visibility by 10%.

LLMs prefer:

• Original statistics and quantitative data
• Expert quotes from credible sources
• Comparison tables and feature breakdowns
• Clear FAQ-style Q&A matching Reddit/Quora training
• Structured data with schema markup

Brand Bias creates compounding advantages for established players. Well-known brands with extensive web presence generate what researchers call “generative brand density” — repeated mentions across authoritative contexts that exponentially increase citation likelihood.

This connects directly to 6sense findings: 85% of buyers had prior experience with the vendors they evaluated — the highest level we’ve recorded. LLMs amplify this incumbent advantage by citing well-documented brands more frequently than emerging competitors, regardless of product quality.

New vendors face a cold-start problem: Limited training data presence equals near-zero visibility in base model knowledge, post-cutoff launches are invisible without web search prompting, and small companies lack the “semantic consistency” LLMs look for across multiple sources.

Category creators capture market value partially because their language and framing become embedded in how LLMs understand and describe the solution space. If you defined the category, LLMs learn and repeat your terminology.

Why buyers don’t trust LLMs alone

These limitations explain why our 2025 Buyer Experience Report shows buyers maintaining 16 vendor interactions despite 94% LLM usage. For $300,000+ decisions, buyers can’t afford to trust AI recommendations that might be:

• Outdated: Base models don’t know about anything after their cutoff date unless prompted to search
• Biased: Training data systematically favors incumbents over better alternatives
• Incomplete: Models miss nuances about pricing, implementation complexity, and AI capabilities
• Unverifiable: Black-box recommendations with unclear sourcing

This verification imperative shows up across market research. TrustRadius found 90% of buyers click through to verify AI-provided information. They’re not blindly trusting LLM output.

The practical implication: LLMs help buyers become better-informed before engaging vendors, but the validation phase remains essential. As 6sense data shows, the point of first contact moved earlier (60% vs. 70% of journey) specifically because buyers need vendor conversations to validate AI-related claims that LLMs can’t reliably answer.

Understanding these dynamics helps explain why account-based platforms that capture multi-signal intelligence become more valuable as research moves into AI-mediated channels. LLMs add opacity, but they don’t change the fundamental need for vendor validation in high-stakes B2B purchases.

Chapter 6: Why 6sense was built for this moment

The panic around LLMs destroying B2B go-to-market strategies misses a crucial point: The dark funnel isn’t new. We’ve been solving for anonymous buyer research since 6sense’s founding. LLMs simply expand what we’ve always addressed.

The problem: when traditional attribution breaks down

Here’s what’s happening across B2B marketing:

According to Search Engine Land’s 2024 analysis, 58.5% of searches now end without a click. Users get their answer directly from AI Overviews or search snippets without visiting any website. Research from Bain & Company shows that when AI Overviews appear in search results, organic click-through rates drop about 30%.

The gap between strategy and measurement has become a crisis. According to 6sense’s own research, 82% of marketers practice account-based marketing, yet only 33% measure account-centric metrics. Organizations are running ABM strategies while measuring individual lead conversions — a fundamental mismatch.

With cookie deprecation, GDPR requirements, and privacy restrictions, only 51% of marketers report being prepared for the cookieless future (Basis Technologies).

The convergence of dark funnel expansion, LLM research adoption, and privacy restrictions creates an unprecedented visibility crisis — exactly the problem 6sense was designed to solve.

6sense Signalverse™: Built to capture what others miss

6sense captures anonymous buyer research through our Signalverse™ technology, which processes over 1 trillion data points daily from B2B companies.

Our multi-signal approach aggregates:

• Anonymous website visitor behavior: IP-level identification showing which companies are visiting your site, what content they consume, and how their engagement evolves over time
• Keyword intent data: Third-party research behavior showing which accounts are actively researching solutions in your category on other sites
• Predictive analytics: AI models identifying high-propensity accounts based on behavior patterns — spotting buying intent before explicit engagement
• Buying committee mapping: Visual persona identification revealing all stakeholders involved in the decision (average 10.1 members per deal), and their level of engagement
• Technographic and firmographic enrichment: Understanding what technology accounts currently use, news happening within companies (such as hiring surges or funding rounds) and whether they fit your ideal customer profile

This account-based intelligence approach captures what individual lead tracking misses. When a buying committee of 10 people conducts research across anonymous website visits, LLM queries, peer conversations, and review sites, you need systems that aggregate signals at the account level to identify buying intent.

The quantified results

6sense customers see measurable improvements across the metrics that matter:

From 6sense’s anonymized customer benchmarks:

• 46% bigger deal sizes: 6sense Qualified Accounts (6QAs) deliver nearly 50% higher deal value than other accounts in customer pipelines.
• 27% faster sales cycles: 6QAs don’t just convert — they close over a quarter faster
• 4x higher conversion rates: When engaging 6sense-prioritized accounts across marketing and sales

According to 6sense’s Sales Intelligence research, only 10% of a B2B company’s TAM is in-market at any given time. By focusing resources on 6QAa, you can boost ROI and growth.

Current state: What we’re seeing

Here’s what 6sense data reveals about the actual LLM impact on B2B buying:

Minimal LLM traffic referrals currently. Despite the panic, traditional search and direct traffic sources remain strong. LLM referrals represent a tiny fraction of overall traffic — growing, but not yet material to pipeline production.

Keyword intent data continues to grow. Third-party research behavior has not been cannibalized by LLM adoption. Buyers still conduct extensive web research that generates trackable intent signals.

Publisher networks unaffected. The display advertising ecosystem, content syndication, and B2B media partnerships continue performing as before. LLMs haven’t disrupted these channels.

Future state: Where we’re going

As LLM adoption matures, 6sense is evolving to capture intent from the sources where buyers AND AI tools conduct research:

Review platform signals: Capturing intent data from G2, TrustRadius, Capterra, and SourceForge — the platforms both buyers and LLMs consult for vendor comparisons. When accounts from your ideal customer profile visit these review sites searching for solutions in your category, that’s a buying intent signal.

Third-party research behavior tracking: Aggregating signals across the broader research ecosystem where buying committees conduct anonymous evaluation.

Enhanced predictive models: Using three years of buyer behavior data to identify patterns that predict buying intent before explicit signals emerge.

Account-level propensity scoring: Weighting multiple weak signals (anonymous website visits + keyword research + review site activity + technographic changes) into strong predictions about which accounts are genuinely in-market.

Why 6sense customers needn’t worry

The core insight from our 2025 Buyer Experience Report should reshape how you think about the LLM impact: 94% of buyers use LLMs, yet maintain 16 vendor interactions.

This proves LLMs complement rather than replace vendor engagement. What matters is being in the consideration set when those 16 interactions begin — exactly what 6sense enables.

Here’s why the dark funnel expansion actually increases the value of account-based intelligence:

When individual tracking erodes, account-level intelligence becomes more valuable. You can’t track every touchpoint, but you can identify which accounts are showing buying intent through aggregated behavioral signals.

Predictive analytics identify opportunities before explicit signals. 6sense models spot high-propensity accounts weeks — sometimes months — before they fill out forms or request demos. This gives your team first-mover advantage.

Multi-signal approach provides redundancy and resilience. If LLM research reduces website traffic, that’s okay. First, serious buyers are still going to visit your website. Second, we capture intent through keyword research, review site activity, technographic changes, and dozens of other signals. No single signal drives our predictions; the aggregate pattern does.

Buying committee mapping reveals engagement across the group. Even if individual committee members conduct LLM research anonymously, we identify the account’s buying intent and map the full stakeholder group.

Most importantly: 77% of buyers in our 2025 data contacted the eventual winner first. Being in that pole position depends on strong brand presence, peer validation, and account-based intelligence identifying when prospects enter buying mode. 6sense helps you achieve all three.

As traditional attribution breaks down and research moves into opaque channels, account-based platforms that can pierce the dark funnel become mandatory infrastructure, not nice-to-have marketing tools.

Chapter 7: Practical strategies: making LLMs your ally

Optimizing for LLM citations ensures you’re included when buyers DO use AI research. These vendor-agnostic tactics complement 6sense intelligence and help build the brand presence that drives consideration set inclusion.

Optimizing for LLM citations (Generative Engine Optimization)

Content formats that boost visibility:

Getting cited by LLMs requires understanding what content formats they favor and where they look for authoritative information.

Research from Princeton’s GEO study provides specific guidance:

• Adding statistics improves visibility 27-36%
• Citing credible sources increases visibility 20-34%
• Including expert quotations boosts visibility 20-35%
• Keyword stuffing actually decreases visibility by 10%

LLMs favor structured data over marketing narratives. Create content with:

• Original research and quantitative findings
• Comparison tables showing features across vendors
• Expert quotes with clear attribution
• FAQ-style Q&A matching natural language queries
• Step-by-step how-to guides with clear instructions

Technical optimization essentials:

Schema markup has become critically important. Microsoft Bing confirmed in March 2025 that schema markup helps its LLMs understand content. Priority implementations include Article schema with author credentials, FAQ schema structuring Q&A content, Organization schema with brand details, and Product schema with features and pricing.

Ensure AI crawlers can access your content by checking that robots.txt doesn’t block GPTBot (OpenAI), CCBot (Common Crawl), Anthropic-AI (Claude), or PerplexityBot. Consider implementing the emerging llms.txt file standard at yourdomain.com/llms.txt to provide structured site summaries specifically for LLMs.

Distribution matters as much as content quality:

Understanding where LLMs find information guides distribution strategy. Research shows Reddit leads LLM citations, followed by Wikipedia, YouTube, LinkedIn, and review sites (Profound).

Publish across high-authority platforms:

• Contribute authentically to Reddit discussions in relevant subreddits. Identify yourself and who you work for, then offer valuable info — not a sales pitch.
• Answer questions comprehensively on Quora (most cited in Google AI Overviews)
• Maintain active review profiles on G2, Capterra, TrustRadius, SourceForge
• Guest post in industry publications and trade journals
• Participate in GitHub discussions for technical products

The timing advantage for early movers:

Content published today on high-authority platforms becomes training data for model versions 6-18 months from now. Unlike daily Google indexing, LLM visibility requires longer positioning horizons but offers more stable returns once established. Organizations building presence now capture disproportionate visibility as models retrain quarterly and annually.

Dominating review sites where buyers and LLMs Look

Our 2025 Buyer Experience Report shows buyers evaluate an average of 5.1 vendors with prior experience on 3.9 of them. Review sites inform consideration sets for both human buyers and AI tools.

The numbers from G2’s research tell the story: Review sites jumped from 13% to 31% of buyers’ most consulted sources between 2021 and 2024 — a 138% increase in three years.

The dual benefit of review presence:

LLMs cite review platforms as “crowd-sourced intelligence” — treating aggregated peer reviews as more authoritative than vendor claims. When ChatGPT recommends software, it often references G2 ratings and TrustRadius reviews.

This creates a reinforcing loop: Strong review presence drives direct buyer discovery AND gets your solution cited when buyers ask AI tools for recommendations.

Systematic review generation:

Implement programs that make leaving reviews frictionless for satisfied customers. Prompt with specific questions about implementation experience, feature usage, comparison to alternatives, quantified ROI, and use-case specific recommendations.

Quality signals matter: A survey by Sopro showed that 77% of buyers place user reviews among their top three most influential resources. According to TrustRadius, buyers looks for detailed reviews addressing pricing transparency (the #1 thing 49% of buyers want), feature breakdowns, implementation complexity, and business outcomes — which provide exactly the comparative information LLMs need for recommendations.

Cross-platform presence creates redundancy:

Don’t rely on a single review site. Build presence across G2, Capterra, TrustRadius, SourceForge, and category-specific platforms. This redundancy ensures visibility even if one platform’s LLM citation frequency changes.

The brand awareness imperative

Our 2025 Buyer Experience Report finding that 85% of buyers have prior experience with evaluated vendors, and 68% already know a seller from the winning vendor validates the strategic shift toward brand-first marketing.

Research from Marketing Week documents this transformation: Brand awareness leapt from sixth to first priority for B2B marketers between 2021 and 2024.

Why the shift matters:

On average, only about 10% of your total addressable market is in-market at any given time. Optimizing conversion funnels merely makes that narrow slice more efficient.

Without brand awareness ensuring you’re in the consideration set when buyers from the 90% out-of-market segment enter buying mode, growth stalls. As our data shows, 77% of buyers contacted the eventual winner first. If you’re not top-of-mind when they start researching, you’ve likely already lost.

The 60/40 budget framework:

LinkedIn recommends allocating 60% of budget to brand building and 40% to demand capture for optimal results — an inversion of traditional B2B marketing that historically allocated 70-80% to demand generation.

The rationale: Brand marketing creates demand that didn’t previously exist and ensures you’re discovered during the 60% of the journey (per our 2025 data) happening before vendor contact. Demand capture tactics only work for buyers already aware and actively searching.

Building consideration set presence:

In an LLM-influenced world where buyers conduct comprehensive research before surfacing, brand awareness determines whether you’re included in that research at all. Execute brand building through:

• Thought leadership content targeting problem-aware buyers
• Category creation and positioning establishing your framing
• Presence on high-authority platforms where LLMs train
• Consistent messaging creating semantic density across channels
• Community building in spaces where target buyers congregate

The 6sense connection: Our predictive models identify which accounts are entering buying mode before they explicitly engage. But they can only engage with vendors they’re aware of. Brand building fills your funnel with accounts that will convert when they hit buying stages 6sense detects.

New measurement for the Dark Funnel era

Traditional metrics — website traffic, search rankings, last-click attribution — increasingly miss what matters. As our research shows, only 3% of website visitors identify themselves through form fills. Measuring the 3% while ignoring the 97% creates a dangerous illusion of understanding.

The 6sense measurement approach:

We focus on metrics that predict B2B revenue:

• Account-level engagement vs. individual lead metrics: Tracking behavior across the entire buying committee, not just form fillers
• Buying stage progression: Understanding how accounts move from Awareness to Consideration to Decision to Purchase
• 6QA (6sense Qualified Account) performance: Our AI-powered qualification, which combines the power of the Singalverse™ with 6AI™ learning models — delivers 4x conversion rates, 46% bigger deals, and 27% faster cycles

Evolving industry standards:

Market data validates this shift. An analysis by Search Engine Land using Ahrefs found a 34.5% decline in clickthrough rates when an AI overview appeared in results. Even top rankings generate less traffic as AI Overviews and LLM usage grow. Average click-through rate for the #1 organic search position dropped from 29% in 2024 to 19% in Q2 2025, according to a study by the SEO agency GrowthSRC.

First-party data as competitive advantage:

Eighty-four percent of marketers are using first-party data strategies to capture customer details, but that data is far less effective when it is siloed in systems that don’t communicate. That makes building Customer Data Platforms that unify behavior across channels mandatory infrastructure.

Organizations effectively using first-party data for personalization can reduce customer acquisition costs, boost revenue, and increase marketing ROI — even while operating on slimmed down budgets, according to McKinsey research.

The strategic implementation focuses on capturing anonymous research on your websites, enriching first-party data with intent signals, building comprehensive Ideal Customer Profiles, creating account-level engagement scoring, and enabling personalized experiences while maintaining privacy.

Self-reported attribution reveals hidden sources:

Simply asking form fillers “How did you hear about us?” reveals sources traditional attribution software misses entirely — podcasts, peer recommendations, AI research, community discussions. This qualitative data provides strategic insights that quantitative tracking can’t capture.

As direct and branded traffic grows while measurable source traffic declines, the pattern suggests influence is working despite traditional metrics failing to reflect it. The HubSpot case study illustrates this perfectly: blog traffic declined sharply in 2023-2024, yet Q4 revenue hit all-time highs.

Brand awareness and authority drive conversions independent of measurable traffic. Measuring influence rather than just trackable touchpoints becomes essential as the dark funnel expands.

Conclusion: The data-driven reality

After three years of research and nearly 10,000 B2B survey responses across global markets, 6sense has documented a consistent pattern: Buyers are confident, independent, and increasingly using AI tools — but they fundamentally still need vendors.

What actually changed

The point of first contact moved from 70% to 60% of journey completion, a meaningful shift representing 3.5 weeks earlier vendor engagement. But the driver isn’t LLM efficiency replacing vendor content. It’s buyers needing to validate AI capabilities in solutions through direct conversations.

94% of buyers now use LLMs during their buying process — one of the fastest enterprise technology adoption rates ever recorded. Yet they maintain the same number of vendor interactions: 16 per person with the winning vendor, statistically identical to prior years.

LLMs are tools buyers use within an existing evaluation process, not replacements for that process.

What stayed remarkably consistent

The fundamentals of B2B buying remain unchanged:

• Buying committee size: 10.1 people conducting extensive independent research
• Vendors evaluated: 5.1, with prior experience on 3.9 (76%) of them
• Winner contacted first: 77% of the time (down from 83%, but still dominant)
• 85% of buyers have prior experience with evaluated vendors — the highest level recorded
• Average deal size: $300,000-$400,000 — too high-stakes to trust AI recommendations alone

These patterns reveal the limits of LLM influence. For complex, high-value B2B purchases requiring committee consensus and vendor validation, AI tools accelerate research but don’t eliminate the need for human relationships and trust-building.

The strategic implications

Three core insights should guide your response to the LLM era:

1. The dark funnel expands, making account-based intelligence more valuable

As individual touchpoints become harder to track and research moves into opaque channels (LLM conversations, private Slack channels, anonymous website visits), account-based platforms that aggregate signals become mandatory infrastructure. You can’t track every touchpoint, but you can identify which accounts show buying intent through behavioral patterns.

6sense customers achieve 4x higher conversion rates, 46% bigger deals, and 27% faster sales cycles precisely because account-based intelligence captures what individual lead tracking misses.

2. Brand awareness determines consideration set inclusion

With 77% of buyers contacting the eventual winner first and 85% having prior experience with evaluated vendors, the battle is won or lost before explicit engagement begins. Traditional demand capture tactics reach only a fraction of the 10% currently in-market. Brand building ensures you’re top-of-mind when the other 90% enter buying mode.

The strategic shift from demand generation to brand awareness isn’t a fad — it’s a mathematical response to how B2B buying actually works. Allocate accordingly.

3. LLMs complement vendor relationships, they don’t replace them

The panic about AI tools destroying B2B go-to-market strategies misses the fundamental reality: Buyers use LLMs to become better-informed before engaging vendors, leading to higher-quality conversations and faster deal velocity. The validation phase remains essential.

Optimize for LLM citations to ensure inclusion when buyers use AI research. Build review site presence to serve both human buyers and AI tools. But recognize that vendor engagement, relationship-building, and trust remain central to B2B revenue creation.

The 6sense advantage

6sense was built for the dark funnel. LLMs simply add another layer to research opacity we’ve always addressed.

In 2024, 6sense customers using intent data and predictive analytics saw big gains when targeting opportunities identified as 6sense Qualified Accounts (6QAs):

• Average opportunity value: 99% higher than non-6QA opportunities
• Average deal value: 46% higher than non-6QA deals
• Win rate: 8% higher close rates
• Time to close: 27% faster sales cycles

These results stem from solving the visibility problem that LLMs intensify: identifying which accounts are in-market before they explicitly engage, mapping buying committees across anonymous research, and predicting propensity to purchase based on aggregated behavioral signals.

As Kerry Cunningham notes in his 2025 Buyer Experience Report analysis, the traffic decline B2B sites experience isn’t from in-market buyers — it’s from the professionally curious and future buyers. Current pipeline production remains visible through multi-signal account-based intelligence.

The path forward

Don’t panic about LLMs destroying your go-to-market strategy. Instead:

Invest in brand awareness that puts you in consideration sets before buyers conduct LLM research. Our data shows 85% have prior experience with evaluated vendors — be one of the vendors they already know.

Optimize for both human and AI discovery through review site presence, LLM citation strategies, and high-authority platform distribution. When buyers ask ChatGPT for recommendations, ensure you’re cited.

Leverage account-based intelligence to pierce the dark funnel and identify buying intent before explicit engagement. The 6sense approach of aggregating signals across buying committees captures what individual lead tracking misses.

Embrace new measurement frameworks focused on account-level engagement, buying stage progression, and influence metrics rather than last-click attribution and traffic volume.

The transformation is real, but it’s not the apocalypse marketers fear. Across three years of research, the core dynamics remain surprisingly stable. Buying committees still make consensus decisions. Vendors still need validation. Trust still matters. Relationships still drive revenue.

Armed with 6sense research and account-based intelligence capabilities, B2B revenue teams can navigate this shift successfully. The competitive advantage belongs to organizations that understand buyer behavior through data and account intelligence.

The future is account-based, intelligence-driven, and brand-aware. The organizations that adapt fastest will capture disproportionate market share as competitors chase yesterday’s playbooks.