Never Split the Difference vs Getting to Yes: negotiation is a pull, not a push

You have prepared properly. You know the numbers, you know the weak points in their position, and you have a clean argument for why your number is the fair one. You make it. It is a good argument. And the person across the table hears none of it, because from the moment you started making it they were composing their reply.

Twenty-five years of doing deals, and this year's formal mediation training at the Straus Institute at Pepperdine, have brought me to an uncomfortable conclusion about my own profession: the thing lawyers are trained to do — advocate, explain, be right — is the thing that most reliably stalls a negotiation. The settlement comes from the other direction.

Two books, two theories of the person across the table

Roger Fisher and William Ury published Getting to Yes out of the Harvard Negotiation Project in 1981, and it has been the standard text since. Its four principles are still the best short course in negotiation anyone has written: separate the people from the problem, focus on interests rather than positions, invent options for mutual gain, and insist on objective criteria. Underneath the four is a model of the counterparty as a reasonable person who, shown a fair standard and a joint problem, will help solve it.

Chris Voss spent his career as the FBI's lead international kidnapping negotiator, and his 2016 book Never Split the Difference opens with him, a working hostage negotiator, sitting in Harvard's negotiation course and finding that the method did not survive contact with his cases. His line is that nobody has ever devised a mutually beneficial win-win solution with a man who thinks he is the messiah. The counterparty in his model is not reasonable. He is frightened, proud, half-listening, and running on feeling — which is to say, he is a person. So the tools change. Voss's method is tactical empathy: mirroring the last few words someone said so they keep talking; labelling their emotion out loud ("it sounds like you feel you have been backed into this"); asking calibrated how and what questions that hand the other side the problem; and treating "no" as the beginning of a conversation rather than the end of one. The goal of the whole apparatus is a specific moment — when the other person says "that's right" — because that is the moment they have felt understood, and nothing moves before it.

Notice what the two books ask you to do with your mouth. Fisher and Ury ask you to propose: standards, options, interests, packages. Voss asks you to receive: reflect, name, ask, wait. One pushes toward agreement. The other pulls the other person toward you until agreement is where they already are.

Why explaining does not work on people

The instinct to explain is not a character flaw; it is wired in. In a 2012 study published in PNAS, Harvard's Diana Tamir and Jason Mitchell found that talking about ourselves activates the brain's dopamine reward system, and that people will give up small amounts of money for the chance to do it. Their estimate is that 30 to 40 per cent of ordinary speech is spent telling other people about our own experience. A negotiation puts two people with that wiring on either side of a table, each waiting for the other to stop so they can get back to the rewarding part.

The person who breaks that loop wins it. In a 2017 study in the Journal of Personality and Social Psychology, Karen Huang and colleagues coded nearly 2,000 four-minute speed dates and found that the daters who asked more follow-up questions — questions that built on what the other person had just said — were more likely to be asked out again. The daters had no idea; the same paper found people consistently fail to predict that asking questions makes them better liked. Substitute "signed the term sheet" for "asked out again" and you have most of what I have learned across the table.

There is a reason the push fails that goes beyond wiring. When you explain, you are doing the other side's thinking for them, and people do not adopt conclusions they did not reach. When you ask "how am I supposed to do that?" — Voss's signature question — you hand them your problem, and a problem someone solves themselves is one they will defend to their own side.

What this looks like in a mediation room

A mediator has no authority at all. The parties can walk out, and the only thing keeping them in the room is that they feel the process is going somewhere. That is why mediation is the purest test of the pull model, and why the numbers are what they are: in the formal evaluation of Ontario's mandatory mediation program, Robert Hann and Carl Baar found 41 per cent of Ottawa cases settled fully at the mediation and another 13 per cent partially, with 85 per cent of litigants reporting a positive effect on cost. Those cases did not settle because the mediator out-argued anyone. They settled because a neutral spent the morning doing what the lawyers could not: asking each side what they actually needed, saying it back to them until they agreed it was right, and then carrying that understanding across the hall.

The research on mediator behaviour says the same thing. Lorig Charkoudian and colleagues sat in on 130 court-connected mediations and found that eliciting — asking the parties what they would propose and testing their reactions to each other's ideas — was the only mediator strategy associated with a higher likelihood of agreement, while directing — explaining the law, offering opinions, advocating an outcome — predicted more adversarial motions in the year that followed. I set out that study in how to choose a good mediator; it is the push-pull distinction with a coding sheet.

Watch what an experienced mediator does. Almost none of it is proposing. It is "help me understand why the timing matters so much," and "so if I have this right, the money is not really the issue, the issue is that you were not told," and then a silence long enough that the client fills it with the thing they had not told their own lawyer. The proposals come late, and by then they are usually the parties' own.

The same technique is available to the parties themselves, and to their counsel, before anyone hires a mediator. The timing question — mediation or litigation, and how early — is one I have written about elsewhere; this piece is about the posture. If you are the lawyer, the hardest part is that pulling feels like not doing your job. You were retained to advocate. Sitting quietly while the other side talks looks, to your own client, like weakness, and it is the opposite: the lawyer who can make the other side feel understood is the one who finds out what they will really take.

Advocacy versus intimacy

I also practise as a relationship counsellor, and the overlap is not a coincidence. In that work the same principle has a blunter name: closeness is a pull, not a push. You cannot make a partner feel close to you by explaining yourself more clearly. You do it by asking, listening, and reflecting back until they feel known. A negotiation is a brief, transactional intimacy between two people who would rather not be in the room, and it runs on exactly the same mechanics. The barriers are the same too: the need to be right, the fear of looking weak, the reflex to answer instead of hear.

Fisher and Ury understood this — "separate the people from the problem" is a relational instruction, and their book asks for active listening as well. The difference with Voss is one of proportion. In Getting to Yes listening is preparation for the real work of reasoning. In Never Split the Difference it is the work, and reasoning is what you permit yourself once the other person has said "that's right." My experience is that the second ordering is correct, and that most deals that fail, fail because someone reversed it.

How to run a pull negotiation

Go in with your numbers and your walk-away point, as Fisher and Ury would insist — the pull model is not softness, and a good BATNA is what lets you afford to listen. Then, for the first stretch of the meeting, do not make a single point. Ask what a good outcome looks like for them and follow up twice. When they object, repeat their last few words with a question mark and wait. When they get heated, name it: "it sounds like this feels unfair." Summarise their position until they say "that's right," and do not proceed until they do. Ask "how would that work on your end?" instead of explaining why it will not work on yours. Treat their first "no" as information about what they are protecting. Make your proposal last, in their language, built from what they told you.

You will talk less than you are used to, and you will leave with more than you expected. It is not that the argument did not matter. It is that nobody was listening to it until they had been heard.