How B2B Buyers Now Use AI in the Research Phase — And What It Means for Your Pipeline

In 2023, a B2B buyer researching CRM software would start with a Google search, visit three or four vendor websites, download a couple of whitepapers, and maybe check G2 before scheduling demo calls.

In 2026, the same buyer opens Claude or Perplexity first.

"I'm evaluating CRM platforms for a 60-person sales team. We use Salesforce today but the contract is up. What are the top three alternatives worth evaluating, and what are the key differences?"

The AI generates a structured response. It names three to five vendors. It explains their positioning. It notes pricing ranges. It often includes a comparison table. The buyer now has a shortlist — before they've visited a single vendor website.

This is the structural change that The Citation Economy is built around. And it has specific, measurable implications for B2B demand generation that most companies haven't fully reckoned with.

The AI Research Phase in Practice

The B2B AI research phase has three distinct query patterns, each corresponding to a different buyer intent:

Category queries: "What is the best [category] software for [use case]?" These queries happen at the top of the funnel — the buyer is still defining their criteria and shortlisting options. This is where initial brand awareness is formed or missed. Brands that appear consistently in AI responses to category queries have a structural advantage in every subsequent buying conversation.

Comparison queries: "[Brand A] vs [Brand B] for [use case]" These happen mid-funnel, when the buyer has narrowed to a shortlist and wants to understand the differences. AI engines draw heavily on G2 Compare, first-party comparison pages, and TrustRadius for these responses. Brands without comparison pages are represented only by third-party sources — often their competitors' comparison content.

Validation queries: "What do users say about [Brand]? What are the common complaints about [Brand]?" These happen late in the buying process, when a buyer is close to a decision and doing final due diligence. AI engines cite G2 reviews, TrustRadius, Reddit discussions, and Gartner Peer Insights here. Brands with thin review profiles or no analyst recognition get vague, unconfident responses — which creates friction at the exact moment the buyer needs reassurance.

What Doesn't Change

The fundamentals of B2B buying haven't changed. Buyers still want:

• Social proof from peers (reviews)
• Third-party validation (analyst recognition)
• Specific evidence of outcomes (case studies with metrics)
• Pricing transparency (to filter before investing in a demo)
• Comparison with known alternatives (to justify the decision)

What's changed is where they look for this information first — and how quickly they form an initial view.

An AI engine can synthesise in thirty seconds what previously took thirty minutes of research. The shortlist that used to form after three days of self-directed research now forms in a single AI conversation.

The Implication for Demand Generation

The traditional demand gen playbook — SEO for awareness, content for nurture, paid search for conversion — still works. But it now operates downstream of an AI layer that most demand gen teams aren't optimising for.

If a buyer has already formed a shortlist via AI before they hit your paid search ad, your ad reaches a buyer who either:

• Already has you on the shortlist (efficient)
• Already decided you're not on the shortlist (wasted spend)
• Hasn't heard of you from the AI (late-stage recovery, harder to convert)

The brands that appear in AI responses to category and comparison queries have already won a form of top-of-funnel awareness before any demand gen spend touches the buyer. The brands that don't appear are fighting from behind — paying to reach buyers who have already formed a view that doesn't include them.

Measuring the AI Funnel

Most B2B marketing teams don't yet measure AI funnel presence. The metrics to track:

AI citation rate: How often does your brand appear in AI responses to your top category and comparison queries? Sample this manually across ChatGPT, Perplexity, Claude, and Google AI Overviews weekly or monthly.

Shortlist conversion: Of buyers who reach your website for the first time, what percentage say they discovered you through an AI tool? Add this to your demo request form or first-call discovery questions.

AI-attributed pipeline: As AI-driven discovery becomes measurable (Perplexity is beginning to provide referral data; others will follow), track the pipeline quality from AI-referred traffic versus organic search.

Brand mention sentiment: Run periodic checks on what AI engines say about your brand in validation queries. Are the common complaints accurate? Are the strengths being accurately represented? Gaps in accurate AI description are a product marketing problem, not just a technical one.

The brands that build this measurement infrastructure now will have a structural lead as AI-driven discovery becomes the default B2B research behaviour. Most competitors aren't tracking AI-sourced pipeline at all yet — the category benchmark is still being set.

---

This post is adapted from Chapters 2 and 3 of The Citation Economy — the playbook for B2B SaaS AI visibility.