94% of B2B Buyers Use an LLM Before They Talk to You. What They Ask About a Person

Buyers run your shortlist past a model for months before a sales call, and part of what they check is you.

The number, and where it came from

6sense published a piece on November 18, 2025 citing its 2025 B2B Buyer Experience Report, a survey of nearly 4,000 B2B buyers. 94 percent of them use large language models during the buying process.

Forrester, cited in that same piece, put LLM use among B2B buyers at essentially zero in January 2024 and 89 percent by mid-2024, with 87 percent of users reporting better business outcomes.

A behavior moved from rare to standard inside about eighteen months.

They arrive later and better briefed

The same 6sense report has buyers contacting vendors at roughly 26.4 weeks into the journey, down from 30 weeks.

Earlier contact came with a stated reason. Nearly 58 percent of buyers who engaged earlier said they needed to validate how vendors had implemented AI capabilities.

Six months of someone else’s research happens before your phone rings. You are being described in that window by sources you did not write.

Sales stopped being the first channel

The 2025 Edelman-LinkedIn B2B Thought Leadership Impact Report surveyed 1,934 global business executives from March 17 to April 3, 2025, with a margin of error of plus or minus 2.0 points. 71 percent of hidden decision-makers report relatively little or no interaction with sales.

Those people shape a shortlist without ever appearing in your CRM. 55 percent of them use thought leadership to vet vendors, alongside 56 percent of target decision-makers.

64 percent trust thought leadership more than marketing materials and product sheets when assessing capabilities. 53 percent agree that if thought leadership is high quality, brand recognition matters less in vetting.

The person-level questions I plan for

No survey I would cite breaks out the exact prompts buyers type about a named individual. What follows is my own working list, built from how buying committees behave, and not a published finding.

Identity comes first: who this person is, what company they run, where they operate, how long they have done the work.

Credibility comes second: whether the person is credible, who else vouches for them, where they have been published or quoted.

Then fit and risk: who to hire for a specific problem, how this person compares to the other two names on the list, and whether anything in the record should give a committee pause.

The answer gets assembled from a short source list

Profound analyzed 680 million citations from August 2024 to June 2025, published June 5, 2025 and updated that August. Within ChatGPT’s top 10 sources, Wikipedia is 47.9 percent and Reddit 11.3 percent, with Forbes at 6.8 percent and G2 at 6.7 percent. Query sampling was not disclosed, so read it as a direction.

Running visibility work at Subscribe PR, the thing that surprises founders is how rarely a model is wrong about them out of malice. It answers from whatever third-party pages exist, and for a lot of founders that set is thin or three years stale.

The traffic from those answers behaves differently

Semrush published a study on July 21, 2025 across more than 500 digital marketing and SEO topics, with prompt and visit counts undisclosed, reporting that an average visit from AI search is 4.4 times as valuable as an average visit from traditional organic search, measured by conversion rate. The underlying conversion rates were not published, so the multiple is the whole claim.

About 50 percent of links in ChatGPT 4o responses point to business or service websites. Commercial pages are already inside the answer layer.

What to do before your next pipeline quarter

Write your version of the four question groups above as literal prompts. Run them on two platforms, paste the answers into a document with the date, and read what the model treats as the record of you.

Then fix supply rather than wording. Add the missing third-party corroboration, update the stale pages you control, and publish the thinking a committee could use to vet you without a call.

If you want the reading laid out for you, the write-ups at calebhellinger.com/insights cover the same ground.

Caleb Hellinger runs Subscribe PR, an answer engine optimization and digital PR agency in Phoenix.

←

Leave a Reply