The aren’t calling anyone. The 6sense 2025 Buyer Experience Report found that buyers make a preliminary vendor choice before ever engaging a seller. And, that pre-engagement favorite goes on to win roughly 80% of deals. 

The same research found that 94% of buyers used large language models during their research for summarizing reviews and analyzing the data they’re pulling together.

Mash those two facts together and you get an unpretty picture. If you’re on the outside of the shortlist, that is. The decision about whether you win is made before first contact, and a machine is now helping make it.

Your marketing plan used to ask, do we rank?

Now, you’ve got to wonder…when the machine summarizes our category, do we get cited? Or, to be more accurate, which of us is getting cited?

Because that’s almost never a brand. When a citation happens, it’s typically of a person. Here’s why, and what you can do about it.

How do B2B buyers actually use AI search?

B2B buyers use AI tools mainly to organize and interpret their research. They summarize reviews, compare approaches, clarify evaluation criteria, and turn a pile of scattered data, like digital tea leaves, into something their buying group can discuss. The machine isn’t necessarily choosing the vendor. But it’s helping shape the case.

That’s a key idea to keep in mind when most of the buying process happens without a seller in the room.

Gartner’s B2B buying research estimates that buyers spend only about 17% of their time meeting with potential suppliers. The rest goes toward independent research, internal discussion, and the exhausting business of getting several people to agree on anything.

The 6sense findings point in the same direction. Buyers spend more of the journey researching than engaging sellers, initiate contact on their own terms, and often reach out only after they have ranked their options.

AI tools now sit inside that research period.

A committee member might ask an answer engine to explain the available approaches, compare competing philosophies, summarize common complaints, or identify the criteria that matter most. The response is assembled from information the system has learned or retrieved.

Some companies make it into that synthesis. Many don’t.

When a company does appear, it’s often thanks to the work of a recognizable and reputable person: an executive with a named framework, a researcher with original data, or an expert who has taken a clear position and supported it.

This is what the first post in this series meant by deal infrastructure. Your executive content is doing work inside conversations you can’t ever attend.

Why do answer engines cite people instead of brands?

Answer engines are more likely to cite people when there’s a clearly identified expert making a specific, supportable claim. But much of traditional brand content gives them very little to work with. The author is vague, the credentials are missing, and the point of view has been softened to the point that it sounds like every other company in the category.

Consider the machine’s problem, then.

A large language model has already encountered an enormous amount of conventional business advice. It doesn’t need another article announcing that customer experience is important, transformation requires leadership, or companies should align sales and marketing.

When a corporate blog repeats familiar advice under a generic byline, the content adds almost nothing distinctive. It may be accurate. It may even be well written.

But there’s no compelling reason to attribute it to anyone. So nobody gets cited.

This kind of consensus content rarely disappears completely. It gets absorbed into the general answer.

What remains easier to identify is material with a clear source: a named person, relevant experience, an original framework, a useful data point, or a claim specific enough that someone else could challenge it.

Search engines have been moving in this direction for years by putting more emphasis on demonstrated experience, expertise, authority, and trust. Those signals are easier to establish when a real person stands behind the content. Answer engines add another layer because a synthesized response often needs sources it can name.

This exposes a weakness in the traditional corporate blog model.

A steady stream of safe, unsigned, committee-approved articles may keep the publishing calendar full. But it doesn’t necessarily give a machine, or a buyer, anything worth remembering. Or, like we keep saying, citing.

How do you get your executives cited by AI search?

It’s straightforward: Give answer engines the basic ingredients attribution depends on, like a named expert with visible credentials, claims that are easy to extract, evidence that can be checked, consistent information about the executive, and corroboration beyond the company website.

1. Put a real name on the work. Each article should identify its author and explain why that person has standing to address the subject. A useful bio describes actual operating experience, not a cloud of adjectives about being passionate, innovative, and results-driven. Author markup should connect the article, the person, and the organization.

2. Make the main claims easy to find. Use headings that reflect the questions buyers ask. Answer those questions directly near the top of each section. Keep paragraphs focused. State definitions and conclusions plainly instead of forcing the reader to excavate them from six paragraphs of throat-clearing.

3. Publish evidence that works outside its original context. Useful evidence includes sourced statistics, documented client outcomes, original research, named frameworks, and concrete examples. These elements can be extracted, understood, and attributed without requiring the machine to reconstruct the whole article.

4. Keep the executive’s identity consistent. Their name, role, biography, and area of expertise should match across the company website, LinkedIn, contributed articles, speaker profiles, and structured data. Small inconsistencies create ambiguity. Clear entity information makes it easier for machines to understand that these references point to the same person.

5. Build evidence beyond your own domain. Owned content is important, but outside validation makes the person and the ideas easier to trust. That may come from industry publications, conference appearances, podcast transcripts, interviews, research citations, or other credible third-party sources. An expert who appears only on the company website is making the machine take the company’s word for it.

Our generative engine optimization guide goes deeper into schema, retrieval, content structure, and measurement. The site-level implementation is part of our Search Visibility work.

None of those mechanics, however, can rescue a point of view that has nothing to say.

The part that’s scary: you need a position

Structure helps a machine understand your ideas. It can’t make those ideas interesting.

If an executive’s published perspective is simply the accepted industry view with a company logo above it, better schema won’t create distinction. It’ll just make the content easier to process before it gets blended into the wider, more beige-y consensus.

A useful position is specific enough to be tested and credible enough to be defended. A knowledgeable person in the field should be able to disagree with it. That does not mean manufacturing controversy or turning the CEO into a LinkedIn carnival barker. It means saying something with consequences.

This is where many thought leadership programs lose their nerve.

Legal removes the sharp edges. Brand broadens the language. Communications replaces the direct claim with something more “balanced.” By the time everyone has approved it, the article is technically flawless and intellectually deceased.

The market is moving in the opposite direction. 2026 research from TopRank Marketing and Ascend2 found that 47% of B2B marketers planned to increase investment in original research and data-driven thought leadership. It also reported that 93% considered research-based content effective for generating engagement and leads.

The formula is not especially mysterious: contribute evidence other people do not have, then use it to support a position people can recognize as yours.

The rest is distribution and formatting.

Does this replace SEO?

No. AI visibility does not replace SEO. The two depend on much of the same underlying work, including structured content, clear authorship, credible evidence, schema, and third-party validation.

Buyers still use search engines. They still visit websites, read articles, compare vendors, and send links to colleagues. The 6sense findings suggest that LLMs are becoming another layer in the research process, not swallowing the entire process whole.

The overlap between search optimization and AI visibility is substantial. Question-based headings help readers and machines understand the page. Named authors establish expertise. Evidence strengthens credibility. Schema clarifies entities and relationships. Third-party mentions support authority.

You do not need two separate content universes. You need a stronger foundation.

What changes is the way you measure the outcome.

Traditional search performance is tracked through rankings, impressions, traffic, and conversions. AI visibility requires a different habit. Run the questions your buyers actually ask through the major answer engines on a regular schedule. Record which people, companies, and sources appear. Note whether your executives are named and whether their ideas are represented accurately.

If they are missing, resist the urge to blame the technology first. The problem may be technical, but it often begins earlier. There may be no distinctive position, no evidence, no consistent author identity, or no outside validation.

Fix the signal before obsessing over the plumbing.

Either way, the shortlist is getting written…with or without you

Your prospects are already using machines to make sense of your market. Those systems are summarizing the category, surfacing sources, and helping buyers decide which options deserve a closer look.

Someone will be represented in those answers.

The people who show up tend to have a few things in common. Their expertise is clear. Their ideas are specific. Their evidence is visible. Their work appears in more than one place. Over time, those signals become easier for both buyers and machines to recognize.

None of that happens by accident. It also doesn’t happen after publishing three generic executive posts and declaring victory before lunch.

The question is straightforward: when an answer engine helps organize the argument behind your next deal, is your company represented by anyone worth citing?

If not, that can be changed. But the work starts with something worth saying.