How to Break a Negotiation Deadlock, and When Not To

The supplier says there is no more movement. Procurement says the increase is still unjustified. The same arguments have already been repeated twice, someone proposes an escalation, and somebody else says the meeting cannot end without a path forward.

That last instinct is understandable and sometimes expensive. A deadlock can be a coordination failure that better questions or a better package will solve, but it can also be the correct commercial outcome when the two sides' minimum conditions simply do not overlap. Treating both situations as the same problem encourages buyers to manufacture movement where no sensible movement exists.

A 2026 review by Martin Schweinsberg, Stefan Thau and Madan Pillutla gives this distinction a useful research frame. Their work treats negotiation impasses as potentially meaningful outcomes and asks how artificial intelligence changes the kinds of deadlock that are likely to remain.

Diagnose the obstacle before searching for movement

Two negotiations can sound equally stuck while requiring opposite responses. In one, the parties misunderstand each other, argue around a single position, fail to generate options or coordinate badly across several issues. In the other, the supplier's true minimum sits above the buyer's maximum, a liability position cannot be accepted, capacity cannot be delivered in the required period or the person in the room lacks authority over the issue that matters.

The practical task is to identify which kind of problem you have. What exactly prevents agreement? Is the disagreement about facts, interests, authority, process or the acceptable commercial range? What new information would actually change either side's position?

Those questions are more useful than asking what concession might get the meeting moving again. Movement has value only when it improves the package, not when it merely removes the discomfort of being stuck.

Many apparent price deadlocks are really package problems

Procurement often meets a “fixed” price position that becomes less fixed when another issue enters the conversation. A supplier may reject a four percent target because headquarters will not approve the margin, yet become able to move when a longer term or better forecast visibility changes the economics. A delivery date may look impossible until the real bottleneck, perhaps qualification capacity at one site, is separated from the rest of the implementation plan.

Face and internal commitments can create a different kind of block. The buyer may already have told finance that the increase is unjustified while the supplier has told senior management that the increase is essential. Even when the economics begin to overlap, neither side wants to reverse itself visibly.

These mechanisms need different responses. Information problems need questions, option problems need trades, authority problems may need escalation, and face problems need a route that allows movement without forcing either side to announce defeat. That is why the ability to change the structure of a negotiation, discussed in our article on adaptive negotiation styles, matters more than loyalty to one favourite tactic.

Face-saving is part of the commercial design

Face-saving is sometimes dismissed as cosmetic, but organizational decisions are carried back to people who were not in the room. A supplier manager may be able to accept a package but unable to report that procurement simply “beat us down,” while a buyer may have room on a term but little room to reverse a public internal position without new justification.

A workable agreement may therefore depend on the story of why movement occurred. The parties can restructure timing, volume, productivity commitments or a review mechanism so that the change is linked to new information or a different package rather than to unilateral surrender.

This is not theatre for its own sake. It is part of designing an agreement that the people in the room can actually get approved and implemented.

AI may remove some deadlocks and make the remaining ones more informative

Schweinsberg, Thau and Pillutla argue that AI can reduce unwanted impasses when the problem is analytical, for example poor option generation, misunderstanding or difficulty coordinating a complex package. A system can compare positions, surface combinations that were overlooked and keep track of many variables without fatigue.

The review also points to a limitation. Humans sometimes reach agreement through aspiration adjustment, relational repair and face-saving, while systems optimized around stable thresholds and narrow objectives may be less flexible on those dimensions. Better analysis can therefore change the composition of the deadlocks that remain rather than eliminate deadlock altogether.

For procurement, that is a useful warning against overinterpreting an optimizer. A system can show that a package sits outside the range it was given; it cannot decide whether preserving a strategic relationship, avoiding an operational disruption or creating a future option justifies changing that range.

A walk-away can be a strong outcome

Procurement often treats agreement itself as evidence of success, which creates pressure to close even when the remaining package is poor. If the supplier's minimum economics genuinely exceed what the buyer can justify, walking away, delaying, changing scope or escalating the business decision may be better than buying agreement through concessions.

The quality of the outcome depends on the diagnosis. “We could not agree” tells the organization very little. A useful conclusion explains what the supplier requires, what the buyer can justify, which trades were tested, why they failed to close the gap and which business decision remains.

That turns an impasse into information. The negotiation may not have produced a contract, but it has reduced uncertainty enough for the organization to decide what to do next.

Practise being stuck, not only breaking through

Deadlocks are difficult to train through theory because stalled conversations change behaviour. People repeat arguments, fill silence, soften positions or escalate simply because nothing is moving. A realistic practice round therefore needs permission to remain unresolved.

Voice2Evolve can create that condition by allowing the simulated counterpart to hold a commercially credible limit instead of manufacturing agreement for the learner. The useful review is then whether the buyer diagnosed the obstacle, tested the sensible paths and knew when continued movement would simply mean paying to avoid an impasse.

A deadlock is not automatically evidence of poor negotiation. It is evidence that the current path has stopped producing movement, and the next skill is understanding why.

Source

• Schweinsberg, M., Thau, S., & Pillutla, M. M. (2026). Negotiation Impasses as Meaningful Outcomes: How AI is Changing the Cost Structure of Walking Away. Current Opinion in Psychology, article 102419.