AI search and AI agents in B2B buying: answers before clicks

The visibility paradox: more research, less visibility

B2B buyers are researching more than ever, and vendors are seeing less of it than ever. AI-assisted research lets buyers compare suppliers before visiting their websites.

Ninety-four per cent of buyers in 6sense’s 2025 B2B Buyer Experience study reported using large language models during their buying process.

AI search visibility means how prominently a brand, its content and its expertise appear in generated answers, summaries and agent-assisted research, beyond traditional search rankings. It is a distinct challenge from keyword optimisation and requires a different response. Buyers ask ChatGPT, Claude or Perplexity a question and receive a synthesised answer, often without clicking through to a vendor website.

This adds to the changes in B2B buyer behaviour: suppliers need to be discoverable during research that may never appear in their website analytics. The response requires organisational infrastructure for managing data, compliance, identity and signal intelligence. Content and keyword changes alone cannot provide those capabilities.

AI in B2B buying now

Imagine a procurement lead at a mid-market SaaS company asking ChatGPT: “What are the best demand generation platforms for enterprise B2B?” The answer can become an initial shortlist before the buyer visits a supplier.

AI adoption in the B2B buyer journey

The adoption figures cover different populations:

• Forrester’s 2025 Buyers’ Journey Survey, a separate study from 6sense’s, reported 94% using AI in their buying process.
• In Magenta Associates’ 2025 UK research, 85% of buyers aged 25 to 34 use AI for supplier research, compared with 23% of those aged 55 to 64.
• 66% of UK senior decision-makers use AI to research or evaluate suppliers, according to Magenta Associates’ 2025 UK research, which surveyed 300 UK B2B purchasing professionals.

AI-assisted research is the default behaviour for the majority of B2B buying teams. The practical question is whether vendors are visible when buyers use it.

Zero-click searches and vendor visibility

AI Overviews can place generated summaries above vendor websites in informational searches. They add to the established pattern of searches that end without a click:

• The zero-click share for AI Overview searches is 83%.
• The broad zero-click estimate is 60%. SparkToro and Datos’ 2024 study found 58.5% in its US desktop and mobile-browser sample.
• The click-through figure for the #1 organic result fell from 0.73 to 0.26, a decline of approximately 64%, following AI Overviews’ introduction.

Visibility and traffic have decoupled: even a page that ranks first on Google can receive fewer visits when the answer appears in an AI summary above the results.

Research vendors cannot see

Unobserved research is research and buying activity invisible to vendors. Peer conversations, analyst briefings and untracked research have always been difficult for vendors to observe. AI tools extend that problem: buyers can compare suppliers in ChatGPT, Claude or specialist systems without generating a visit or form submission.

For example, an agent may synthesise dozens of sources into a recommendation that the buyer uses without visiting the recommended supplier. The supplier’s analytics cannot reconstruct that entire interaction.

Receiving answers before visiting a supplier

Buyers receive synthesised answers before clicking through to vendor sites, if they click through at all. This is “answers before clicks”: a structural change in how buyers gather information, evaluate options and shortlist vendors, beyond zero-click search alone. In this AI-mediated research environment, visibility means being cited in the answer rather than simply ranking for a keyword.

GEO means generative engine optimisation. The GEO study by Aggarwal and colleagues, including Princeton researchers, found that adding statistics improved visibility in its experiments. AI models prefer data-rich, clearly structured content. The brands that provide it get cited. The brands that rely on generic marketing copy do not.

Why AI search visibility matters for vendors

A buyer may form an initial view of a supplier from an AI answer. If the supplier is absent from the answer, the buyer may never encounter it during that research.

Concentration of citations

Research attributed to BrightEdge and Amsive in 2025 reports an average of 3–4 brands per response, with the top 20 domains receiving 66% of citations. That creates a winner-takes-all pattern: a few brands dominate the answer while the rest are effectively invisible within it.

10Fold’s 2025 “AI-First, Buyer-Ready” study surveyed 400 senior marketing executives at B2B technology companies. Only 11% reported having 75 to 100% of their content optimised for AI search.

The gap between brands visible to AI and those absent from its answers is vast and widening. Traditional search can still send traffic to a supplier ranked fifteenth. Citation in an individual AI summary is closer to binary: the brand is mentioned or it receives nothing from that answer.

Conversion from AI referrals

Early figures attributed to Averi.ai in 2026 suggest that AI referrals can convert at higher rates than organic Google visits:

The figures attributed to Averi.ai give multipliers of 5.6× to 6.0×. These early figures need broader validation. They reflect the nature of AI-referred traffic: buyers arrive with context and a comparison already made. That makes an AI citation arguably more valuable per visit than a first-place Google ranking.

Citation authority and search visibility

Citation authority means being mentioned or quoted in AI-generated answers. Its three drivers are data presence, entity prominence and statistical content: making information available, making the organisation unambiguous and providing specific evidence. That gives suppliers a visibility objective beyond accumulating backlinks.

For Google AI Overviews and AI Mode, Google’s guidance says SEO fundamentals remain relevant and no special AI markup is required.

The future: agent-intermediated commerce

What is agentic commerce?

Agentic commerce means B2B transactions intermediated by AI agents, from research and vendor evaluation through negotiation and contract execution within the agent’s authority.

Gartner’s 2025 forecast puts more than $15 trillion of B2B spending through AI agent exchanges by 2028, with 90% of B2B buying intermediated by agents. It does not establish a universal timetable for autonomous purchasing.

Partial adoption is happening now: buyers use agents for research and shortlisting while human decision-makers still sign contracts. Infrastructure requirements compound as agents move from shortlisting to negotiation and purchasing. At every stage, agents need structured, current, machine-readable data about products, terms and evidence.

Implications for vendor visibility

Make decision information accessible. Brand narratives matter less than the availability of pricing, specifications, compliance evidence and performance data.

Check the sources agents can reach. Agents may use websites, search results, knowledge bases or other data sources. Information missing from those sources is less likely to inform their evaluation.

Expect incomplete measurement. Some agent-assisted research produces visits and referrals; some leaves no signal in the supplier’s analytics. Citation frequency, data availability and repeated checks of agent responses can complement visits, click-through rates and marketing-qualified leads (MQLs).

Data infrastructure for agent visibility

The records that support agent visibility cover pricing, compliance and performance evidence.

Performance evidence includes case study metrics, processing volumes and accuracy rates. Describe compliance evidence accurately: ISO 27001 certification, SOC 2 reports and information about GDPR compliance are different kinds of assurance, not interchangeable certifications.

Maintaining this information requires systematic data governance, structured markup and institutional commitment. Clear ownership, review dates and consistent product records help maintain it across sources.

Why AI visibility needs organisational infrastructure

Tactical and infrastructure thinking

Tactical thinking asks, “How do I optimise my content for AI search?” It focuses on individual pages, keywords and formatting. Those responses are necessary but insufficient.

Infrastructure thinking asks, “What systems do I need to become visible to AI agents?” It treats data, compliance, identity and signal intelligence as foundations. AI visibility is an organisational capability that cannot be built through a blog post or keyword strategy alone.

Foundations to maintain

• Compliance by design: AI agents must be able to trust the data. Build GDPR compliance, AI governance and legal transparency into the information systems.
• Signal intelligence: monitor, test and iterate to understand which signals agents seek when evaluating vendors.
• Lawful identity activation: make data available where agents research, including structured databases, knowledge bases and AI-indexed content, with the appropriate consent and compliance controls.
• Systematic data management: assign responsibility for keeping product, company and evidence records accurate and current.

These foundations work together as one system for demand generation, rather than separate projects or quarterly campaigns. We call this the Demand OS approach.

What delay can cost

Competitive displacement. With 3–4 brands per response and the top 20 domains receiving 66% of citations, competitors are claiming the available mentions. Absence means losing a place in the buyer’s consideration.

A gap in reaching younger buyers. The 85% AI usage among buyers aged 25 to 34, against 23% among those aged 55 to 64, points to a generational shift. Brands invisible during AI-assisted research are invisible to those buyers during that part of their decision.

Infrastructure debt compounds. Building clean, structured, machine-readable and compliant data infrastructure takes six to twelve months of sustained effort. Every quarter of delay gives competitors another quarter to build their foundations.

From clicks to answers and from tactics to infrastructure

Visibility in AI-mediated buying depends on maintaining data, compliance, identity and signal intelligence together as organisational capabilities. Keyword improvements and additional blog posts cannot, on their own, build those foundations.

For the structural shifts behind this trend, see B2B buying behaviour. Defining the ideal customer profile (ICP), buying group and anti-target identifies whose questions the information needs to answer.