The middle of the funnel did not disappear. It moved into AI systems, and most B2B vendors have no presence there. Buyers still evaluate vendors, compare capabilities, and build business cases before making a purchase decision. They just do all of that work through Perplexity, ChatGPT, and Claude instead of through calls with your sales engineers and SDRs.
This is about the middle: the stage where buyers move from “I know this category exists” to “I’m ready to talk to two specific vendors.” That stage used to belong to vendors. It now belongs to AI.
What the Middle of the Funnel Used to Look Like
The traditional middle of the funnel was vendor-controlled. To understand pricing, you called sales. To see a demo, you booked a call. To get a comparison, you downloaded a gated guide. The vendor controlled the evaluation experience because the vendor controlled access to evaluation information.
This created a useful asymmetry. You could sequence information disclosure to maximize perceived value. The vendors who owned the narrowing process, who guided a buyer from five options to two, won the deal.
What Happened to It
AI eliminated the information asymmetry that made vendor-controlled evaluation necessary. A CFO can now ask Perplexity to compare two ERP vendors on features, pricing, implementation timeline, integration capabilities, and customer satisfaction. They get a synthesized answer in 30 seconds that would have taken three sales calls and two demos to construct.
The evaluation conversation that used to happen with a sales engineer now happens with an AI system. The shortlist that used to be built through vendor outreach now gets built through AI-generated vendor shortlists. The business case that used to require a vendor’s ROI calculator now gets built through AI synthesis of public information, review data, and community discussions.
The buyer never fills out a form. They never book a call. They complete the entire middle-of-funnel evaluation process through a conversational interface that has no awareness of your attribution model.
What This Does to Revenue Teams
The SDR’s outbound sequence is reaching buyers who have already completed their evaluation. The prospect already knows your capabilities, your pricing range, your competitive positioning. The SDR is not creating curiosity. They are interrupting a decision that was already made.
The demo is no longer the moment of evaluation. It is the moment of confirmation. Buyers arrive knowing what they expect to see, having already formed a preference through AI research. If a buyer has already decided through their AI-driven buyer journey that your platform cannot meet their needs, a great demo will not change their mind.
Marketing’s attribution model shows a direct visit or a branded search, but the actual journey started with an AI query days earlier. The website visit is not the start of the journey. It is the end of the middle-of-funnel research phase that happened entirely inside an AI interface.
The New Middle of the Funnel
The new middle of the funnel is the set of AI-mediated interactions where buyers move from category awareness to vendor preference. It happens in the dark funnel, outside your attribution model, before buyers identify themselves. This activity generates no MQLs, no demo bookings, no trackable engagement.
Buyers run dozens of queries during evaluation. They start broad: “What are the best platforms for this use case?” They get more specific: “How does Vendor A compare to Vendor B on this feature?” Each query is an evaluation moment. Each answer shapes preference. By the time a buyer surfaces in your CRM, they have already done the work your middle-of-funnel process was designed to guide them through.
The buyers are not following your buyer’s journey map. They are following their own question sequence, guided by what matters to them and what AI surfaces as relevant.
What the Winning Vendors Are Doing
The vendors winning this shift have presence in the AI systems that now run evaluation conversations. When a buyer asks AI to compare them to a competitor, the AI has accurate, detailed, current information about their capabilities. These vendors understand what the AI Demand Channel is and have built strategies to own it.
They have built evaluation content as a strategic priority: technical documentation, comparison frameworks, pricing context, and implementation details that AI systems synthesize into evaluation answers. Content structured for AI agents is specific, citable, and comprehensive, not vague summaries or marketing copy.
They have third-party validation across the sources AI systems cite: review platforms, analyst reports, community discussions, industry publications. They measure middle-of-funnel performance by Share of LLM, not demo bookings.
What to Do About It
Start by auditing your middle-of-funnel AI presence. Run an AI channel audit: run the queries your buyers run and see what AI says about you. Compare your product to your top competitor. Ask about pricing, implementation timelines, integration capabilities. If AI cannot answer these questions accurately, your buyers are getting incomplete or incorrect information during the evaluation stage that now happens before you ever hear from them.
Build structured evaluation content as a strategic priority. Pricing pages, integration guides, implementation timelines, competitive comparison frameworks. This content used to live in sales decks and demo scripts. It now needs to live in formats AI systems can extract, synthesize, and cite.
Stop measuring middle-of-funnel performance by demo bookings and MQL volume. These metrics measure the moment a buyer surfaces in your CRM, not the moment they started evaluating you. Track citation quality instead: inclusion rate, position, accuracy, and source diversity across the evaluation queries your buyers run, and decide how often you need to re-check that data based on how fast your category moves.
The middle of the funnel still exists. It just exists outside your CRM, outside your attribution model, and outside your ability to influence it through traditional sales and marketing motions. The companies that adapt will own the new middle. The ones that do not will continue optimizing for a buyer journey their prospects have already abandoned.
For the full strategy on building presence in the AI-mediated middle of the funnel, see How to Build an AI Demand Channel and A6 Group’s AI Channel Strategy services.
Related Content
What Is the AI Demand Channel — The definitive guide to the channel where B2B buyers now research and evaluate vendors before ever contacting sales.
What Is Prospect Experience? — The totality of what buyers encounter during their invisible research phase, and why it determines shortlist membership.
How B2B Buyers Use AI to Build Vendor Shortlists — How buyers construct shortlists through AI research before vendors know they exist.
Content for AI Agents — How to structure content so AI systems can extract, synthesize, and cite it during buyer evaluation.
AI Channel Audit — How to audit what AI systems say about your company during the evaluation stage.
The AI Buyer Journey — How buyers move from awareness to decision using AI systems instead of vendor-controlled touchpoints.
Evaluation Content — Why evaluation content, not awareness content, is the highest-value asset for AI-mediated buyer research.
Your Website Is Selling Without You — Why your website is now running evaluation conversations with AI agents that you never see.
Share of LLM — The metric that measures your presence in AI-generated answers during buyer evaluation.
Dark Funnel — The buyer research and evaluation activity that happens outside your attribution model and CRM.
Continuous AI Testing vs Periodic Audits — When monthly checks are enough and when the middle-of-funnel categories that drift fastest need more frequent testing.
The middle of the B2B funnel is the stage where buyers move from category awareness to vendor preference. This stage used to be vendor-controlled: buyers had to engage with sales engineers, attend demos, and download gated content to get evaluation information. AI eliminated that information asymmetry. Buyers now complete the entire evaluation phase through Perplexity, ChatGPT, and Claude without ever contacting a vendor. The middle of the funnel still exists. It moved into AI systems.
SDRs now reach buyers who have already completed their evaluation through AI research. The prospect already knows your capabilities, pricing range, and competitive positioning before the first outreach. SDRs are not creating curiosity. They are interrupting a decision that was already made. Sales engineers face a similar shift: the technical questions they used to answer on discovery calls are now answered by AI systems before the discovery call exists. Their role has shifted from education to confirmation.
Buyers who have researched through AI arrive at demos with pre-formed preferences. They know roughly what your product does, what it costs, and how it compares to competitors. The demo is no longer the moment of evaluation. It is the moment of confirmation. If AI research created an inaccurate impression of your capabilities, the demo either confirms it incorrectly or creates a disconnect between what the buyer expected and what they see. Fixing the AI research phase is now upstream of fixing the demo.
Demo bookings and MQL volume measure the moment a buyer surfaces in your CRM, not the moment they started evaluating you. The evaluation now happens before that surface moment. Middle-of-funnel metrics should track citation quality in AI systems: inclusion rate (how often you appear when buyers run evaluation queries), position (where you appear relative to competitors), citation accuracy (whether AI describes you correctly), and source diversity (whether citations come from owned content and third-party sources). Share of LLM across evaluation queries is the north star metric for the new middle of the funnel.
Vendors need evaluation content: technical documentation, comparison frameworks, pricing context, integration details, and implementation timelines structured for AI extraction. This content used to live in sales decks and demo scripts. It now needs to live in publicly accessible formats that AI systems can read, synthesize, and cite. Vague marketing copy and gated resources are invisible to AI. Specific, structured, ungated evaluation content is what AI systems surface when buyers run evaluation queries.