G2 introduced new buyer-intent tools this year as B2B software research continues moving into AI chatbots. Its 2026 Buyer Behavior Report found that 82% of B2B buyers sourced software recommendations from an AI chatbot in the past two years, while an earlier G2 study, The Answer Economy, found 51% now start their software research with an AI chatbot more often than Google.

AI can help build a vendor shortlist before someone fills out a form. It can narrow a category before anyone reaches a website, moving a meaningful part of the research process outside the marketer’s view. For demand-gen teams, that makes the beginning of the funnel much harder to see.

AI Search Is Moving B2B Research Earlier

Most demand-gen systems were built around observable behavior. A person visits the website, downloads something, attends a webinar, clicks an ad or fills out a form. Those actions create signals that marketing can score, attribute and eventually pass to sales.

Increasingly, none of that is the first thing to happen. A buyer can work through much of the evaluation inside an AI chatbot before a single one of those signals fires.

That doesn't mean websites, search or sales calls disappear. The Answer Economy report found that 71% of B2B software buyers use AI chatbots somewhere in their research, while many continue using traditional search alongside them. The difference is that more consideration can happen before the company has any direct interaction with the buyer.

HubSpot found something similar in its own research. Among more than 3,000 CRM purchase decision-makers, buyers using AI search were 36% more likely to purchase than those who didn't. At the same time, HubSpot reported declining organic website traffic across nearly 300,000 businesses on its platform.

For demand-gen teams, that makes it harder to see what buyers are doing before they reach the website. A buyer can become highly informed and fairly close to a decision while still looking like an anonymous prospect from the company's point of view.

Buyer Intent Can Exist Before a Lead

Traditional funnel reporting often treats the first identifiable action as the beginning of demand. In reality, that action may now happen much later.

Someone who requests a demo might already know the major vendors, understand the pricing models, have read customer complaints and eliminated two competitors. A form fill may come only after the buyer has already done most of their research.

Some familiar demand-gen metrics become harder to interpret. A decline in organic traffic or content downloads doesn't automatically mean interest has fallen if buyers are getting answers elsewhere. A lead that converts unusually quickly may have already spent weeks researching before the company could track any of it.

Pipeline metrics only show part of what’s happening, so demand-gen teams need to look earlier in the research process too.

Demand Generation Has to Move Upstream

Teams need to know which questions buyers are asking early in their research and which sources they rely on for answers.

That could mean identifying the comparisons people make early in research, the concerns that repeatedly delay a purchase, the communities where buyers discuss the category and the third-party sources they trust when evaluating vendors.

G2's Answer Economy report also found that 69% of surveyed buyers ultimately selected a different software vendor than they initially expected, and 33% purchased from a company they hadn't previously heard of after AI surfaced it. That suggests early consideration is becoming more fluid, not less.

Those early research habits also affect what companies need to put into market. Customer reviews need to answer useful questions. Educational content needs to address the way buyers actually compare options. Credible third parties, communities and creators can matter because AI systems pull from information that exists well beyond a company's own pages.

Audience Intelligence Can Start Before Lead Capture

This is where studying an audience’s micro-communities, behaviors, interests and influences becomes useful for more than targeting. It can show marketers what people are researching, which conversations are shaping their decisions and where those conversations happen.

That information can shape the subjects a company explains, the proof it puts into market and where it needs to be credible. It's a problem Lickly, one of our portfolio companies, is built to work on.

The Form Fill Is No Longer the Beginning

Demand gen has spent years getting better at what happens after someone raises their hand. AI search makes the period before that moment harder to ignore.

The first measurable interaction may now arrive after a buyer has already learned the category, compared vendors and formed an opinion. Teams that only study the visible funnel risk optimizing the final part of a decision they barely influenced.

Demand gen now has to account for more of what happens before a buyer becomes a lead, including the information shaping the shortlist and the places buyers go to find it.