Research Is the New First Call

The first call happens without you

6sense runs one of the largest recurring studies of B2B buying behavior, close to 4,000 buyer responses in the 2025 edition. Fair warning before the numbers: 6sense sells software built on the premise that buyers decide early, so they have skin in this conclusion. I discount the decimals and read the mechanics, and the mechanics are hard to dismiss. Walk through what the study found, in the order a deal unfolds.

Buyers are roughly 61 percent of the way through their buying process before they contact a seller. Seventy-nine percent of the time, the buyer initiates that first contact, not the rep. Before that contact happens, 94 percent of buying groups have already put their vendor shortlist in order of preference. When the list is ranked like that, the vendor they reach out to first goes on to win the deal about 80 percent of the time. When it is not, that number falls to 57. Hold onto that gap. It is the size of the opening you still have.

Read that sequence again slowly, because it describes a sales call that has already been held. The research got done. The candidates got ranked. A winner got penciled in. Then, and only then, your phone rang. Ninety-five percent of purchases in the study went to a vendor on the buyer's day-one list, the names they had in mind when the process started.

1. 01
   A question
2. 02
   AI answers
3. 03
   Quiet site visits
4. 04
   Peer proof
5. 05
   Ranked shortlist
6. 06
   First contact

That chain is the actual first call now. Every link in it happens without a seller in the room.

Where the research moved

The research layer is my plain-English name for everywhere a buyer forms an opinion about you before contact: the AI engines, your website, review sites, and their peers. The newest numbers show how much weight the AI engines now carry in it.

All four numbers are from G2's March 2026 survey. G2 has its own stake here, they run a review marketplace, which is why I care less about any single decimal than about the direction, and the direction matches what Gartner heard from the 646 buyers it surveyed in late 2025.

The line worth sitting up for is the shortlist one. On the 6sense numbers, the shortlist all but decides the deal, and buyers now say AI chatbots influence it more than any other source. The answer a buyer gets from ChatGPT at 9pm on a Tuesday is doing work your best rep used to do.

The reshuffle is real

Two more G2 findings, and these are the ones I would put in front of a leadership team.

Sixty-nine percent of buyers said an AI chatbot surfaced information that led them to pick a different vendor than the one they originally planned to buy. And 33 percent bought from a vendor they had never heard of before an AI engine recommended it.

The second one deserves a minute. A third of buyers gave money to a company that was not in their head when they started. Set that beside 6sense's finding that 95 percent of purchases come off the day-one list and the two seem to fight. Different studies, different buyers, different questions, so I will not chain the arithmetic. But there is a reading that fits both, and I think it is the right one: the day-one list still wins, and the engines now help write it. The vendor a buyer “always knew about” is, more and more, a vendor they first met inside an answer. Which means a well-positioned smaller firm can now walk into shortlists that brand recognition alone used to lock, and an established one can quietly fall out of answers it assumed it owned.

Nobody is grandfathered in. That cuts both ways. The firms I spend my days with are the smaller, sharper kind, and this is the first shift in years that tilts the table their way.

One more beat, for everyone whose business runs on referrals, which describes most of the businesses I serve. A referral used to be the whole first call. Now it is the seed of one. The name your happy client passes along gets typed into an engine and checked against everything else in the research layer before the call gets returned. The referral opens the door. The research decides whether you walk through it.

The twist in the newest data

Here is the part the doom headlines skip, and it is my favorite finding in all of this research.

Buyers are talking to sellers earlier again. 6sense's 2025 read puts first contact at 61 percent of the way through the journey, down from 69 percent the year before, roughly six to seven weeks sooner. One reason: 58 percent of buyers said they engaged sellers earlier specifically to evaluate vendors' AI capabilities, the kind of thing a chatbot cannot demo for them. And buyers using LLMs still average 16 interactions with vendors across a purchase. The conversations did not go away.

So the research layer did not delete the seller. It changed the entry fee. The buyer arrives having done real homework, and the reward for surviving their research is an earlier, longer, more serious conversation than you were getting three years ago. The engines start the funnel. They cannot finish it.

What the first call is for now

Which brings me back to the title. The buyer's research is the new first call: it is where the questions your discovery script used to ask are now asked and answered, alone, at their pace.

The call you eventually get is a validation call. The buyer arrives with requirements drafted and a preference penciled in, and they are checking it: does the story hold up live, do the claims survive questions, do they trust you at the seams where the research could not reach. Running a classic discovery script on that buyer, the fifteen questions about their business you could have answered yourself, tells them the research was one-sided. They studied you. You did not study them.

Treat validation as the honor it is. You made the shortlist. The buyer is handing you a chance to confirm what the research layer said about you, and to add the things it never can: specifics for their situation, honest tradeoffs, proof on demand.

Four moves, in order

1. Ask the engines about yourself. Open ChatGPT, Perplexity, and Gemini and ask what a buyer would ask: best options in your category, your company by name, you versus the competitor you lose to most. Read the answers like a buyer. It costs nothing and takes an afternoon.
2. Fix what the answers get wrong at the source. The engines assemble answers from what they can find and cite: your pages, your listings, third-party coverage. Thin or missing coverage reads as a verdict. Where the answer is wrong or empty, trace it to the source and fix it there.
3. Feed the trust layer. In G2's data, 45 percent of buyers said review-site citations inside an AI answer are the most confidence-inspiring signal in it. Reviews, named proof, and verifiable claims have become load-bearing. The engines lean on them when deciding whether to vouch for you.
4. Rebuild the first call as a validation call. Retire the interrogation-style discovery script. Assume the buyer has read everything, open by finding out what they already believe, and spend the call confirming, correcting, and adding what the research could not give them.

The first two moves are exactly where an AI Search Power-Up starts: what the engines actually say about you, graded, and the fixes in priority order.

The call you cannot skip

The first call is not gone. It moved upstream, into the answers and pages a buyer reads alone, and it is held whether or not you show up prepared. The research layer is where deals are won now. The validation call is where they are kept. Both are yours to lose, and only one of them comes with an invitation.

Research is the new first call. Win the one you are not invited to.

All signal. No noise.