We don’t need to research how AI chatbots have disrupted B2B discovery. We’ve lived it. Starting a new company put us on a B2B shopping spree: CRM, ATS, AI visibility reporting, consent management. A whole Christmas list of software. Whether it was a CRM we were fairly sure we’d pick regardless (HubSpot) or a consent management platform we researched from scratch (Termly), AI chatbots played a role in every purchase.

When we started our last agency in 2012, we spent a minimum of four to six weeks on every B2B purchase, sitting through demos, combing review sites, asking peers and calling references. This time, AI chatbots compared pricing and features for us, sized fit for our stage (Greenhouse was more than we needed for an ATS), checked integration compatibility (Framer worked with most of the website martech we needed) and stress-tested preferences we already held (going with Peec.ai for AI visibility reporting).

In almost every case, AI shaped the shortlist. Your buyers are doing the same: Forrester found 94% of business buyers use AI in their buying process.

For the vendors we didn’t select, the double whammy is that not only did they not get our business, but their missed opportunity was invisible. The consent management platforms that didn’t get surfaced in a response and the ATS vendors that Claude dissuaded us from considering didn’t know they were in the running. No site visit to attribute, no demo request, no sales call. They lost a deal they never saw.

How has B2B research changed?

The shortlist used to form before the research started. Top-of-funnel ads, events, analyst quadrants and peers put vendors on it. The research path that followed could be exhaustive and exhausting: G2/Capterra, Reddit and forums, industry pubs, analyst reports, YouTube demos. All of that to trim or validate a list the buyer already had.

Highly researched, highly customizable, often expensive purchases are exactly where a synthesized answer pays off most, adding both efficiency and context. Instead of Google ranking links across YouTube, G2/Capterra, Reddit and industry pubs, ChatGPT weighs the videos, articles and posts from those same sources and turns them into comparison tables, summaries and recommendations.

So it’s no surprise that G2’s 2026 survey found AI chatbots influence which vendors make buyer shortlists more than any other source, review sites included.

The research still happens. The model just does the reading.

Do B2B buyers rely on AI more than consumers?

Knowledge industry workers are often utilizing new technology first. B2B organizations are adopting AI faster than consumers are. Forrester found business buyers are twice as likely as consumers to use ChatGPT and four times as likely to use Microsoft Copilot.

Usage is one thing. Buying is another. High price tags and low risk tolerance should make B2B buying groups more skeptical of AI, and they are: TrustRadius found that 94% of buyers fact-check AI answers.

However, it’s important to differentiate between AI buying and AI research and discovery. We’re not talking about agentic commerce and letting AI purchase on your behalf. It’s about formulating the shortlist and validating the choice you do make. “Nobody ever got fired for buying IBM” was about making a safe consensus choice. LLMs are the ultimate consensus machine. Picking the vendor the model recommends feels about as safe as it gets.

Does AI shorten the B2B purchase cycle?

This one is unclear. 6sense found average cycle length dropped by more than 10% in 2025, from 11.3 months to 10.1, while Forrester found that purchases including genAI features stretched decision timelines by 30%. Any read on the full purchase cycle carries a lot of noise: economic uncertainty, changing buying committees and the fear of committing to software that AI itself might quickly make obsolete.

The more useful lens is the part of the cycle marketing can actually influence, and there the data is consistent: discovery and research are contracting. In the same G2 survey, over 80% of buyers said AI chatbots made their purchase decisions faster, and 6sense found time spent evaluating each vendor fell from 2.6 months to 2.0. It passes the sniff test. Reviewing a vendor’s website, reading a handful of Capterra reviews, watching a YouTube demo or two and wading through a few Reddit threads takes well over an hour per vendor. A thorough five-page summary of all of it from Claude takes a slow reader fifteen minutes.

AI referral traffic also converts better. Ahrefs found AI referrals made up 0.5% of its traffic but drove 12.1% of signups, and Semrush found AI search visitors convert at 4.4 times the rate of traditional organic visitors. A 50-client study we ran at our last agency confirmed the pattern at a far more modest size: conversion rates 10-20% higher. Either way, B2B buyers are arriving more educated and readier to start a sales conversation.

The cycle may not have gotten shorter, but the part you can influence did.

Does this make advertising more or less important?

When we bought single sign-on software for the first time at Metric Theory, Okta made the shortlist partly because of the mindshare its ads had built across SF and online. If we had Gemini back then, we likely would’ve used its shortlist instead.

There’s a logical argument that if buyers are just going to rely on an AI-provided shortlist, then pumping in awareness spend that used to go toward earning your spot on that same shortlist is moot. On the flip side, it’s a safe bet that HubSpot would’ve been on our CRM shortlist agnostic of whether it was included by AI. This is a product of years of awareness, advertising, content marketing, word of mouth and hands-on utilization.

G2 found that 33% of buyers still purchased from a vendor they’d never previously heard of. We believe advertising still plays an important, albeit changed, role. There are three main jobs to be done by upper-funnel advertising in the age of AI research:

1. Get your brand named in the prompt. Buyers often ask AI to compare X vs. Y vs. Z using vendors they already know, and TrustRadius found 79% of buyers had heard of the product before they started researching.

2. Break ties within the AI shortlist. Consensus machines hedge: AI answers are far more likely to recommend five vendors than to crown one. Being the brand the buyer recognizes tilts the scale when you’re one of five.

3. Feed the flywheel. This one is our working theory: more advertising leads to more customers, which leads to more reviews, discussion and coverage, the raw material AI recommendations are built from.

Whether that nets out to more spend or less is still an open question. Either way, advertising’s job is now getting you named in the prompt and chosen from the answer.

What does this mean for owned media moving forward?

B2B owned media is starting to face the scrutiny paid media has lived with for nearly two decades. The squeakiest wheel in a B2B org is almost always the top salesperson, and the complaint that gets the grease will sound like this: “[Insert tier 1 prospect] said we didn’t come up in ChatGPT when they searched for top [application category].”

Content, SEO and review programs will increasingly have to prove their impact on pipeline. At Aftergrowth, the metric we hold them to is Recommendation Rate: how often AI chatbots actually recommend your brand, rather than just mention it. It’s the most predictive measure of whether you’ll make the shortlist. Remember the CMP and ATS vendors that lost our business without knowing it. That loss was invisible, but it didn’t need to be inevitable.