G2: Half of B2B Buyers Now Start with AI. 69% Switch Vendors Because of It.

The G2 2026 AI Search Insight Report surveyed 1,076 B2B software buyers in March 2026. AI is now the majority-first step in software research. Not trending toward it. There.

The 51% Crossing

In April 2025, 29% of B2B software buyers started their research with an AI chatbot. Twelve months later, 51% do.

The category now has more AI-first buyers than Google-first buyers. The B2B marketing stack built on search queries, comparison pages, and nurture funnels now describes a minority of new purchases.

This happened fast. A company that built its brand visibility strategy in 2024 built it for a buyer journey that no longer describes the majority. Companies that moved early on AI citation infrastructure — G2 presence, structured data, comparison pages, llms.txt — are compounding now. The ones that waited are watching competitors appear in responses where they don't show up at all.

The 69% Redirect Problem

Nearly 7 in 10 buyers switched from their intended vendor because of what AI told them.

From the perspective of the losing brand: buyers enter the market with your company in mind, then leave with someone else. This isn't lost awareness. It's pipeline that existed and got redirected mid-decision. The buyer knew your name. They opened ChatGPT or Perplexity to validate. The AI pointed them somewhere else.

This is why AEO is different from traditional brand marketing. You can run strong top-of-funnel awareness, hold organic rankings, and appear in analyst reports, then still lose the deal because the AI the buyer consulted at the moment of decision didn't include you — or included you poorly. For 69% of buyers, what AI says is what sticks.

The 33% Opportunity

One third of buyers purchased from a brand they had never heard of before an AI recommended it.

AI has no memory of last quarter's LinkedIn spend. It doesn't favour category incumbents. It recommends whoever it can cite with confidence, based on structured, verifiable evidence it can point to.

For a company with 200 G2 reviews, a well-structured comparison page, and a clean llms.txt file, this is a real growth lever. A buyer who has never heard of you opens Perplexity, asks "what's the best [category] tool for [use case]," and gets your name — not because you outspent anyone, but because you built citation infrastructure that made you recommendable. The problem is that recommendation only happens if you've done that work.

What Determines Who AI Picks

The G2 data tells you that AI is now the majority research channel. The practitioner question is what determines which vendor gets recommended.

Based on audit data across hundreds of B2B SaaS domains, six signals drive most of the variance in AI recommendation rates:

Citation Readiness: G2 profile, third-party review volume, analyst mentions, structured product data. AI engines cite what they can verify from sources they trust.

Brand Authority: Domain authority, inbound link profile, how often the brand appears as a named source in industry content. Higher authority means more retrieved results.

Content Quality: Structured content that directly answers the evaluation questions buyers ask. Comparison pages, ROI frameworks, use-case specifics. AI retrieves and cites content that matches query intent.

Platform Optimisation: Schema markup for SoftwareApplication, AggregateRating, and FAQ. Structured data that makes product claims machine-readable.

Technical AEO: llms.txt, robot directives for AI crawlers, server-side rendering that makes content accessible to GPTBot, PerplexityBot, and ClaudeBot.

Competitive Positioning: How the brand appears in direct comparison queries ("vs", "alternative to", "competitors"). These carry the highest buyer intent. Brands with structured comparison content tend to dominate them.

What to Do This Week

1. Run a free audit. Before optimising anything, see where you currently stand across ChatGPT, Perplexity, Gemini, and Claude. Most teams are surprised by the gap between perceived brand strength and actual AI citation rate.

2. Check your comparison page coverage. "[Your product] vs [competitor]" queries are the highest-intent prompts AI receives. If you don't have structured, AI-readable comparison pages for your top 5 competitors, you're conceding the evaluation phase to whoever does.

3. Verify your G2 review count. The 51% of buyers starting with AI are using G2 as a primary trust signal. Under 50 reviews and you're rarely cited. 500+ and you're cited consistently. If you're between those thresholds, a 90-day review programme is probably your highest-ROI action right now.

4. Add llms.txt. A one-page file at your root domain that tells AI crawlers who you are, what you do, and what content to prioritise. It takes an hour to implement and a few weeks to take effect.

5. Audit competitor comparison pages. If competitors are being recommended over you in AI evaluation queries, it's usually because they have structured comparison content and you don't.

---

The 51% figure is a milestone. The 69% figure is the operational problem. Buyers are entering the market ready to buy from you and leaving with someone else because AI redirected them at the moment they were closest to converting.

See where you rank across ChatGPT, Perplexity, Gemini, and Copilot — free audit, no sign-up required.

---

Data: G2 2026 AI Search Insight Report, n=1,076 B2B software buyers, March 2026. AI opinion quotes are generated responses to standardised prompts, not editorial positions.