No Starts the Negotiation
Chris Voss spent fifteen years negotiating hostage cases for the FBI. His method is tactical empathy: label the feeling, invite the no, summarize until they say that is right, and never split the difference.
The Core Insight
Two Harvard professors demanded 1,000,000 dollars for Chris Voss's son in a classroom role-play. One of them, Robert Mnookin, directs the Harvard Negotiation Research Project. Voss brought no case. He asked how he was supposed to do that, then asked for proof the boy was alive. Three minutes of that, and they folded.
Voss spent fifteen years as an FBI hostage negotiator, and he led international kidnapping cases among about 10,000 agents. Never Split the Difference, written with Tahl Raz in 2016, is that job turned into a method. Negotiation gathers information and changes behavior, and both jobs run on emotion.
Most business negotiators treat the table as a rational exchange between logical actors, with feeling as noise around the deal. Voss treats feeling as the machinery. His term is tactical empathy. Understand the mindset of the other side in the moment, hear what sits behind it, and say it out loud.
The bureau tested the split on itself. Lanceley and Noesner asked 35 experienced officers how many had resolved a crisis through classic bargaining. Not one hand went up. Every hand went up for emotional crises with no clear demands.
The Framework
The spine is a five-stage ladder the FBI calls the Behavioral Change Stairway Model. It comes from Carl Rogers and unconditional positive regard. Each stage pays for the next one.
- Active listening comes first, because nothing after it works on a person who feels unheard.
- Empathy means naming the other side's perspective well enough for them to recognize it.
- Rapport arrives after the naming, and it is trust rather than friendliness.
- Influence becomes available once rapport exists, and never before.
- Behavioral change is the output, and the only proof the earlier stages worked.
Most negotiators try to start at influence. The two stages in front of it are the whole price of getting there.
Six tactics do the first stage: effective pauses, minimal encouragers, mirroring, labeling, paraphrase, and summary. Paraphrasing plus labeling equals a summary. Say the meaning back, name the feeling under it, then stop talking.
Key Ideas
Mirrors and Labels Are the Cheapest Moves
Mirroring repeats the last three words the other person said, or the one to three that carry the weight. It copies words alone, never the accent or the posture. Then it needs silence. Richard Wiseman tested it on waiters, and the mirrorers averaged tips 70 percent higher than the rest.
Labeling names the feeling you observe. A label opens one of three ways: it seems like, it sounds like, it looks like. The word I puts a guard up, so a label never starts with it. The last rule of labeling is silence. Lieberman at UCLA found that naming an emotion moves activity out of the amygdala and into the areas that reason.
The accusation audit lists every terrible thing they can say about you, then says it first. A contract paid Anna's firm for nine people, then five and a half, then three. She opened the renegotiation by naming their view of her as the small business being pushed out. Anna's firm left with 1,000,000 dollars.
A No Opens the Table
Yes is three animals, and only one of them ends in a signature.
- The counterfeit yes is an escape hatch for someone who plans to say no.
- The confirmation yes is a reflex to a black-or-white question, and it promises nothing.
- The commitment yes is the real one, and it ends with a signature on a contract.
A no is a statement of perception more often than a statement of fact. Voss lists seven meanings, from not ready to agree, to needing more information, to wanting to consult someone else. Two urges sit under all of them: the need to feel safe and the need to feel in control.
So the openers change. A cold call asks whether now is a bad time to talk, because a no there costs nothing. An ignored thread ends with one email asking whether they have given up on the project. It works on the fear of loss and implies you are walking away. Ben Ottenhoff rebuilt a fundraising script around no-oriented questions and got a 23 percent better rate of return.
That Is Right Is the Signal That Counts
In August 2000 Abu Sayyaf took Jeffrey Schilling, twenty-four, near Jolo Island and priced him at 10,000,000 dollars. His mother had access to perhaps 10,000. The kidnapper Sabaya built his number from figures he had heard elsewhere, including a reported 20,000,000 dollars from Gaddafi for six Western European captives. He called the demand war damages.
Talks ran four months. Voss stopped arguing about money and summarized the world according to Sabaya back to him, war damages included. Sabaya went quiet for nearly a minute, said that is right, and never mentioned money again.
That is right marks the moment the other side weighs your summary and calls it correct on their own. You are right is what people say to make someone go away, and it buys about a day of peace. Voss ran it on his son Brandon, who told his coaches they were right and kept smashing into blockers. One label named the belief underneath, that dodging a block was unmanly. Brandon said that is right, and St. Thomas More won every game.
Questions Hand the Other Side the Wheel
A calibrated question has no fixed answer and cannot be closed with a yes. It buys time, hands over the illusion of control, and makes the other side solve your problem. Drop can, is, are, do, and does from the front. Keep what and how, and use why only where defensiveness serves you.
The best one is a single line. How am I supposed to do that? It creates forced empathy, since they have to solve the problem they made. Rebels seized Julie's husband Jose on the Mira River in August 2003. The captors opened at 5,000,000 dollars. She answered every demand with a question and drove the ransom to 16,500. Their negotiator kept walking back to camp because he had no authority to answer.
A yes is worth nothing without a how. Two implementation questions check it: how we will know we are on track, and how we will handle being off track. A promise to try is a plan to fail. The people who kill deals outrank the people who make them.
Mehrabian at UCLA split a message into 7 percent words, 38 percent tone of voice, and 55 percent face and body. When they disagree, label the mismatch out loud. The rule of three then takes the same agreement three times in one conversation, through three different doors. Take the commitment, then a summary that earns a that is right, then a question about slipping.
Anchors Bend What Fair Means
Kahneman and Tversky published prospect theory in 1979, and two findings run the bargaining chapters. Certainty carries a premium, so people take a sure 9,499 dollars over a 95 percent chance at 10,000. Flip the same numbers into losses and they gamble. A loss stings at least twice as much as an equal gain.
Six moves bend the number, starting with their emotions. Voss ran an accusation audit about looking like a lousy businessman, then offered subcontractors 500 dollars a day against a usual rate of 2,000. All of them took it, and none countered. Let the other side go first in most cases.
Name a range, because a job paying between 130,000 and 170,000 dollars makes 130,000 look cheap. Pivot to what costs them nothing, the way a magazine cover story bought a discount on a 25,000 dollar day rate. Use odd numbers, since a figure ending in zero reads as a placeholder. Then add a surprise gift, which triggers reciprocity.
The Ackerman model is the one fully specified procedure in the book, and it comes from Mike Ackerman, a former CIA officer.
- Set the target price first, and treat it as the goal rather than a hope.
- Open at 65 percent of that target.
- Plan three raises of decreasing size, to 85, 95, and 100 percent of the target.
- Use empathy and several ways of saying no to make them counter before each raise.
- Make the final number precise and nonround, closer to 37,893 dollars than to 38,000.
- Add a nonmonetary item they probably do not want, to show you are at the limit.
The increments halve each time, which convinces the other side they squeezed you to breaking. People given a sequence of concessions feel better about the process than people handed one firm fair offer. Voss ran the ladder in Haiti, where victims priced between 15,000 and 75,000 dollars. One family met a 150,000 dollar demand for an aunt. The nephew anchored at 3,000 dollars and closed at 4,751, plus a portable CD stereo nobody asked for.
Three Unknowns Sit on Every Side of the Table
Voss works from one hypothesis. Each side of a negotiation holds at least three pieces of information that change everything, and neither side knows it holds them. He takes the name from Taleb and calls them black swans.
On June 17, 1981, William Griffin took nine bank employees hostage in Rochester and shot a teller on his own deadline. Three signals had sat in plain sight. The bank kept hanging up, and hostage-takers always talk because they always have demands. A double homicide two blocks away went unconnected. A note read aloud mentioned the police taking his life, and nobody read that far. Griffin wanted suicide by cop.
Pull comes in three kinds, defined as the ability to inflict loss and withhold gain. The positive kind is whatever they said they want. The negative kind is the ability to make them suffer, and it runs on loss aversion. Naming it out loud beats using it. The normative kind uses their own stated rules, because nobody likes to look like a hypocrite.
The hunt works on a spreadsheet too. A student negotiating a Charleston building listed at 4,300,000 dollars ran one deliberate mislabel about the seller's doubts. The answer named the reason for the sale, troubled properties in Atlanta and Savannah. He closed at 3,600,000.
Practical Applications
Write the one sheet before the conversation, since Voss puts the return on preparation at seven to one. Write only your best case as the goal, because obsessing over a walk-away number turns it into your target. Add the summary that earns a that is right, plus three to five labels and calibrated questions.
Run the accusation audit out loud in the first two minutes. Denying an accusation gives it credit, and saying it yourself buys the rest of the meeting.
Change the openers. Ask whether now is a bad time to talk. Send the one-line email asking whether they have given up on the project. Mislabel on purpose when a counterpart goes silent, because the correction is the engagement.
Price with the ladder. Open at 65 percent of your target, make them counter before each raise, and keep the final number odd. Then take the same commitment three times before you call it closed.
Who This Is For
Founders who negotiate face to face get the most from it: fundraising, enterprise sales, hiring, and any contract with a stronger party. Skip the book if you price at volume and nobody ever talks.
The gap between the book and a founder's table is real, and Voss never closes it. Hostage negotiation is a one-shot game against a counterpart you never see again. A founder negotiates with an investor who joins the board and a first hire who reads the terms again next year. Some of the moves read as manipulation in that setting. A deliberate mislabel is a small lie. An accusation audit run on a co-founder is theater. The parts that transfer carry nothing hidden: labeling, summarizing until you earn a that is right, calibrated questions, an odd final number. Leave in the book anything you cannot explain out loud to the person across the table.
The evidence is thin in places. War stories fill more than half the pages, and most chapters spend one on a single rule. The student wins repeat one shape: prepared labels, a calibrated question, an odd number, a win. The bargaining personality types rest on a single study of lawyers. Figures come from the author's own cases with no independent audit.
The Decision
Take the smallest negotiation on your calendar this week, where losing costs you nothing. Remove the counteroffer from your first ten minutes. Prepare five labels, and open by saying the worst thing the other side can say about you.
The test is one phrase. Count how many times you hear that is right before either side names a number. Zero means you were arguing and the summary never landed. Two or more means the price conversation starts somewhere other than the middle.