Buyers are shortlisting before your first touch. Your pipeline plan should start there too.

Most pipeline plans begin at the first touch you can track, such as a form fill, a reply, or a booked meeting. The people arguing loudest about AI search say the shortlist is decided before any of that.

LinkedIn’s Mimi Turner wrote in June that “94% of buying groups use Large Language Models (LLMs) like ChatGPT or Gemini before talking to sales,” citing 6sense’s 2025 buyer experience report. In the same post she points to research LinkedIn ran with Bain & Company finding that brands are 20 times more likely to be bought when everyone in the buying group knows them from the start. The post doesn’t give sample sizes or methods for either figure, and 6sense sells software built on this premise, so we treat both numbers as leads.

This week Toby Coulthard, chief product and growth officer at Jacquard, made a stronger claim in Crunchbase News: “Often, in fact, AI agents will make purchasing decisions before a human ever sees the product page.” He also says Gmail and Apple now summarize emails before the recipient opens them. Jacquard sells AI-generated marketing copy, so he has a stake in this too.

Our take

We think the evidence is thin, but the planning gap is real either way. If even part of the shortlist forms in an AI answer, a plan built on traffic and click-through rates starts measuring after the decision that matters most. TechInformed made the same argument in September: discovery has moved into channels marketing can’t see, and campaign planning hasn’t caught up.

A new “AI visibility” dashboard won’t close that gap on its own. Those tools report what a model said to the prompts you chose to track, and they can’t tell you which prompts your buyers actually used. Your buyers know, and the cheapest way to find out is to ask them.

What to change this quarter

• Add one discovery question: “Which vendors were on your list before you talked to us, and where did the list come from?” Log the answer in a field on the opportunity so you can report on it.
• Split win rate by shortlist status. Compare deals where you were on the buyer’s original list with deals where you got added later. If the gap is large, the front of the funnel deserves budget your attribution model isn’t giving it.
• Read your outbound the way the inbox will show it. If Gmail or Apple Mail puts a summary first, that summary is your opening line. Send a test sequence to a seeded inbox and see what it says.
• Once a month, ask ChatGPT, Gemini, and Claude the questions your buyers ask in discovery, and note whether you show up and whether the description is right.

What would change our mind

We’d drop this if shortlist data from a few dozen deals showed buyers still found vendors mostly through peers, events, and search clicks we already track, or if win rates didn’t differ by shortlist status. We’d also take the 94% figure more seriously if 6sense published its sample and question wording. If you have either, send it to us.

Have you asked recent buyers where their shortlist came from? What did they say?