Two Buyers Researching Together Is Worse Than One. The Peer-Reviewed Mechanism.

The recap call felt like the best one of the quarter. Two champions on the line, not one. They had read everything. They had run the comparison through some AI tool and come back with a tidy grid. They finished each other's questions. They agreed with each other, fast, and they agreed about you.

You hung up sure the deal was won. Two engaged buyers, aligned, doing their homework. What is better than that?

Nothing happened. Three weeks later, the deal was a no-decision.

Here is the part that should bother you. That call was not the moment the deal was healthiest. It was the moment it was most likely to die, and the thing that killed it was the exact thing you read as a buy signal: two people, confident together.

The wisdom the whole playbook is built on

Every sales methodology treats a bigger, more engaged buying group as good news. More stakeholders means more diligence. More voices means fewer blind spots. Two heads are better than one. So when a committee shows up prepared and unified, the rep marks the deal up, the forecast firms, and everyone moves on to the next call.

The buying groups have certainly gotten bigger. Recent buyer research puts a typical B2B purchase at thirteen internal stakeholders plus nine external influencers (Forrester, The State of Business Buying, 2026). Gartner sizes buying groups at five to sixteen people across as many as four functions (Gartner Sales Survey, 2025).

The assumption underneath all of it is that more people researching together produces a better decision. The decision science says that assumption is backwards.

The peer-reviewed mechanism

Start with the individual. Research in Decision Support Systems found that a more elaborate explanation aid raises a person's decision confidence while reducing the cognitive effort they spend and producing a worse choice (Shi et al., 2011). Read that twice. The better the aid looked, the more certain the buyer felt, and the worse they chose. Confidence and quality moved in opposite directions.

Now add the second person. A 2021 study found that participants working in pairs were more overconfident than those completing the same task alone (Plonsky et al., 2021). Not better calibrated. More overconfident. Pairing did not cancel out individual error. It inflated the shared certainty around it.

Put the two findings together and you get the mechanism. Two buyers researching together do not average out each other's mistakes. They validate each other's conclusions and walk away more sure than either of them had any right to be. The group does not catch the error. The group ratifies it.

One honest boundary, because it matters. These studies were run in consumer-choice settings, and a direct B2B vendor-selection replication has not been published. The mechanism is well established. The exact magnitude inside a buying committee is still an open question. I would rather tell you that than pretend the number is nailed down. But the direction is not in doubt, and it is the opposite of what your pipeline assumes.

AI pours fuel on the fire

Now hand that pair an AI research tool, which is what every buyer is doing in 2026.

A 2026 synthesis of fifty peer-reviewed studies found that AI and LLM assistance reliably raises a decision-maker's confidence without improving their accuracy, opening what researchers call a calibration gap: the distance between how sure someone feels and how right they actually are (Consensus Deep Search, 2026). Longer, more polished AI explanations widen the gap. The output looks more authoritative, so the buyer trusts it more, while the accuracy underneath does not move.

It gets worse on the decisions that matter most. Research in IEEE Transactions on Human-Machine Systems found that giving a decision-maker more AI-generated information and options raises their confidence while degrading the outcome, and the damage is largest on exactly the hard, high-stakes decisions (Ward, 2023). A vendor selection that will shape the next three years is precisely the kind of hard decision where the AI assist hurts the most.

Same honest boundary applies. This evidence is about AI-assisted decision-makers in general, not B2B buyers specifically. It gives the mechanism peer-reviewed backing. It does not yet prove the committee-level magnitude. The claim is a well-evidenced mechanism with B2B application still inferential, and anyone who tells you otherwise is selling something.

So the bigger the group, the bigger the gap

Stack the three layers and the picture is clear.

One buyer with a slick decision aid gets more confident and less accurate. Two buyers with that aid get more confident than either of them alone. A committee of eight, each arriving with their own AI-assembled certainty, spends the meeting confirming what the tools already told them. Nobody is the brake. Everybody is the accelerator.

That is why the recap call felt so good. The unison you heard was not the sound of a well-vetted decision. It was the sound of a group converging on a shared conclusion and amplifying its confidence in it. The committee did the diligence the playbook asked for. The diligence made it surer, not righter. And a confidently wrong buying group does not argue with you. It quietly decides you are not worth the risk and stops returning calls.

This is the hidden engine under the no-decision loss. More than half of qualified deals that die end in no decision rather than a competitor win (Dixon and McKenna, Harvard Business Review, 2022). Some of those are buyers who could not get to yes. But some are buyers who got to a confident no, together, and never told you the reasoning because the reasoning felt obvious to them.

The turn

So name the thing you have been doing.

You have been scoring confidence as readiness.

Engaged committee, aligned answers, prepared questions, a unified front. You read all of it as the deal getting healthier, when half of it was the deal getting more certain of a conclusion you never got to see, let alone correct. The signal you trusted most was the one most likely to be manufactured. Confidence is not readiness. It never was. It is just the feeling readiness is supposed to come with, and a group with AI tools can generate that feeling without any of the substance underneath it.

What groups actually get right

This is not an argument that buying groups are bad or that you should wish for a single decision-maker. The research cuts both ways, and the honest version includes the other half.

Gartner found that when a buying group experiences what it calls group relevance, the alignment of shared interests over individual priorities, the group is three times more likely to report a high-quality deal. Aligned groups can outperform individuals. The same Gartner work also found that seventy-four percent of B2B buyer teams show unhealthy conflict during the decision. So the spread is enormous. A genuinely aligned group is your best outcome. A group that has manufactured the appearance of alignment around shared AI research is your worst, and from the outside, on a recap call, the two can sound identical.

The failure mode is specific. It is not the group. It is the group, plus a decision aid that inflates confidence, minus anyone whose job is to check the calibration. Remove that third element and the group is an asset. Leave it missing and the group becomes a confidence machine pointed at a conclusion nobody stress-tested.

That is the gap a readiness diagnostic closes. It does not ask whether the committee is engaged. Engagement was never the problem. It asks whether the committee's confidence is earned, and where the decision is actually stuck underneath the certainty.

Three signals of a confidently wrong buying group

You can spot this in the field before it costs you the deal. Three signals, in order of how often they show up.

1. Unison without dissent. A healthy group can name the strongest case against its own conclusion. A miscalibrated one cannot, because nobody in it ever played that role. If you ask "what is the best argument for not doing this," and the committee goes quiet or gives you a strawman, you are looking at agreement that was never tested, only shared.

2. Certainty that outruns the evidence. Ask the group to show its work. A calibrated buyer walks you through the reasoning. A miscalibrated one points at a conclusion and an AI summary nobody pressure-tested. When the confidence is high but the workings are thin or borrowed, the confidence is the tool's, not theirs.

3. Confidence that rises as the stakes rise. This is the tell that matters most. Real calibration gets more tentative as a decision gets harder, because the person can feel the uncertainty growing. Manufactured calibration gets louder. If the committee sounds more sure about the biggest, hardest part of the decision than the small parts, that is the Ward effect in your pipeline: the AI assist failing hardest exactly where it is trusted most.

When you see these, the move is not to celebrate the engaged committee. It is to diagnose where the decision is actually stuck and recalibrate the group before its confident conclusion hardens. You become the third element the group is missing: the one checking the calibration instead of adding to the confidence.

Back to the recap call

Picture that call again. Two champions, prepared, aligned, finishing each other's sentences. The grid. The fast agreement. The feeling that you had won.

You had not won. You had watched a pair do exactly what the research predicts a pair does: build shared confidence faster than shared accuracy, with a tool that widened the gap between the two. The call was not the green light. It was the last moment you could have stepped in and asked the question that would have surfaced what they were confidently getting wrong.

Two heads are not better than one when both heads are nodding at the same screen.

The next time a buying group sounds this unified, this sure, this done, do not mark the deal up. Ask what they are certain about, and ask them to prove it. If you want a structured way to score whether a committee's confidence is earned across all four readiness dimensions, the four-minute buyer readiness assessment is built for exactly that.