How AI Search Is Reshaping B2B Lead Funnels

Forrester's 2026 research found that 94% of B2B buyers used AI somewhere in their most recent purchase³. For a service business or a growing software company, this means the shortlist is often drawn up in a chat window. That happens well before anyone fills in a form or books a sales call. The funnel still exists, but its early stages have moved somewhere most companies can't see. That changes what a lead looks like when it finally arrives, how you work out where it came from, and what your website is for.

Quick Summary

AI search results are reshaping B2B lead funnels by moving vendor comparison into chat answers, before a buyer contacts any company. Forrester's 2026 research found 94% of B2B buyers used AI in their latest purchase, so the website now does its work at a later checking stage. That means judging content by the inquiries and revenue it produces, and keeping your public description consistent enough for an assistant to repeat accurately.

Most Vendor Comparison Now Happens Before Anyone Contacts You

Most B2B buyers now compare vendors and build their internal case with an AI assistant before they get in touch with a single company. In Forrester's research, 55% of buyers used AI to compare vendors and 47% used it to build an internal business case, both ahead of any direct contact³.

That second number deserves a moment. A business case is the document someone writes to get a purchase approved by their boss or their finance team. If a buyer drafts it with an AI assistant before speaking to a vendor, they've already settled on what the problem costs, which kind of solution fits and roughly what they expect to pay. By the time they reach you, a fair amount of that thinking is written down and has been shared inside their company. AI search lets buyers explore a category, understand use cases and weigh common problems without speaking to sales¹. That early, do-it-yourself stretch of the B2B buying journey is the part that has grown.

These figures have limits. "Used AI during the purchase" can mean anything from one question typed into a chatbot to a full vendor shortlist built there. The numbers also reach us through a summary of Forrester's work, and the full report may break them down differently. The shift matters less for some businesses, too. If most of your work comes by referral in a local market, the buyer usually has your name before they search for anything, and an AI answer mostly confirms what a colleague already told them.

The Six Steps a Buyer Takes, From First Question to Signed Deal

The B2B funnel now tends to run in six steps: a question, an AI answer, a shortlist, checking against other sources, a visit to the website, and then the purchase³. The older journey was shorter and more direct. People searched, clicked, read several vendor sites and then got in touch.

The most striking thing about the new order is where the website lands. It comes fifth, after the buyer has heard about you from an AI tool and gone looking for someone else's opinion. Two things follow from that. Your own site is less often the place where someone first learns what your category is. And the outside places where you're mentioned, such as reviews, industry write-ups, partner listings and forum threads, carry more weight, because that's where buyers go to test what the AI told them.

Why Organic Traffic Can Fall While Demand Holds Steady

A drop in organic traffic often means buyers are getting their early answers in a chat window, where they used to click through to your blog⁵. Demand for what you sell can hold steady through it. Innovaxis, a B2B marketing firm, points out that researching a complex business problem used to mean reading many articles, case studies and reviews across several companies before reaching out. AI-assisted search squeezes that process into far fewer steps⁵. Their advice is to work out what's declining, and why, before cutting a content budget.

Picture a firm whose site gets 5,000 organic visits a month. About 3,500 of them land on explainer posts like "what is a client portal", and the other 1,500 land on its service, pricing and case study pages. If AI answers take over the explainer questions, the first group might halve while the second barely moves. Total traffic would fall by around a third, which looks alarming on a dashboard. Yet the people who were ever likely to become leads are still arriving.

A useful way to read your own numbers is to split them into three groups and look at the same months side by side:

• Explainer traffic: visits to "what is" and how-to pages, the content most exposed to AI answers.
• Decision-page traffic: visits to services, pricing, comparisons and case studies.
• Inquiries and closed deals: the outcomes the other two are supposed to feed.

Suppose the first group falls while the other two hold. Then the site is mostly losing early-stage readers who were only browsing anyway. If decision-page traffic and inquiries drop together, something else is probably going on, and it's worth digging further before putting it down to AI.

What Makes an AI Assistant Name Your Company?

Nobody outside the AI companies knows exactly how their assistants weigh sources. What we do know is that the assistants build answers from what's publicly written about a business. Clear, consistent public information about what you do, who you do it for and roughly what it costs gives them something accurate to repeat. Imagine your website says one thing, an old directory listing says another, and a review site describes a service you dropped years ago. The answer a buyer gets is likely to be muddled or out of date.

Being named seems to count for a lot. GEO SEO Lab reports that 85% of buyers view a vendor more favourably when an AI system names it specifically³. The summary doesn't say who ran that survey, though, so the precise figure deserves some caution.

Where your description lives

• Your homepage and service pages, especially the opening lines of each
• Directory and marketplace listings, including ones set up years ago and forgotten
• Review sites, and the words customers use to describe you there
• Articles, podcasts and partner pages that mention you by name

Most of the work here is tidying. It means making sure these sources say the same few things in plain words, with current services and a current idea of price.

A quick check you can run this week: ask two or three AI assistants the question your buyer would type, such as "which firms help a 20-person accounting practice move client files into one portal?" Note whether you're named, how you're described, and whether any price range mentioned matches what you charge today.

Your Website Now Does Its Work at the Checking Stage

Because the website comes late in the journey, its main job is to help a buyer confirm a choice they've half made. That means answering shortlist questions quickly: who you're a good fit for, what the work involves, roughly what it costs and what happens after they get in touch. A buyer arriving from an AI answer already has a rough picture of your category. What they want to find out is whether the AI's picture of you was right, and whether you suit their situation.

In practice, specific pages tend to do best here. Take a service page that names who it's for in its first two lines, say a 15-person design studio or a regional logistics firm. It lets a buyer rule themselves in within seconds. Case studies that name the industry and give real before-and-after detail supply the kind of proof the buyer was hunting for at the checking step. A pricing page, or at least a page explaining how pricing works, answers the question their business case already raised. Teams reworking service pages for buyer fit often start with those opening lines, because an assistant tends to lift the same sentences a skimming reader does.

Leads Now Arrive Knowing Your Competitors' Names

When an AI-assisted buyer finally gets in touch, they usually know who else they're considering and have a view on what the problem costs. They may also be carrying a few things the AI got wrong about you. So the first sales conversation moves away from introducing the category and toward helping them weigh up their options.

That puts more weight on judgment. An AI assistant can gather and summarize options quickly. It can't tell whether this buyer's timeline is realistic, whether the person championing the purchase has the budget authority they think they have, or whether the problem they've described is the real one. Those are the calls an experienced salesperson or founder makes on a first conversation. They're worth more now that the buyer has done the basic research on their own.

It helps a great deal to know what the buyer already believes before that call. Many teams now add a few lead qualification questions to their inquiry form. These might ask which other options the buyer is looking at, what prompted the search and what they've read about you so far. Splitting those questions across a few short steps tends to feel lighter to the person filling them in than one long form does. AMW Funnels builds inquiry forms that way, and each answer is saved with the new lead. So whoever takes the call can see beforehand that the prospect is weighing two other firms and was quoted an old price by a chatbot.

How to Measure a Funnel You Can Only Partly See

The most dependable measure is revenue by source, backed up by asking buyers directly how they found you. A lot of AI-assisted research leaves no trail in web analytics. Someone who hears your name in a chat answer may later type in your web address or search for your company name. That visit often shows up as direct or branded traffic, with no hint of where the interest started.

Three habits help fill in the gaps:

1. Add a free-text "How did you first hear about us?" field to your inquiry forms. Answers like "ChatGPT suggested you" or "a colleague mentioned you" are imperfect, but they tell you things analytics can't.
2. Track branded searches and direct visits over several months. If they rise steadily while explainer traffic falls, discovery may be happening somewhere you can't see.
3. Tie closed deals back to where they came from, so each channel is judged on the money it brings in.

For that third habit, AMW Attribution matches each closed deal to the source that brought it in. That lets you see whether a fall in blog visits was followed by less revenue from organic search, or whether paying customers kept arriving by another route. When the self-reported answers and the revenue picture point the same way, you can make budget decisions with far more confidence than a traffic chart alone would give you.

If your inquiry form is still a single box asking for a name and email, it's a good place to start. AMW Funnels shows how a short, multi-step version collects what a buyer already knows before your first call.