Ruinous Empathy Costs the Most

The Core Insight

Kim Scott called a new hire's work a good start. The work was incoherent, a kind of word salad. She said a version of that for ten months.

She liked him and did not want to make him cry. She assumed his earlier form was about to come back. Fixing the work herself was faster. Colleagues covered, deadlines slipped, and results decayed.

Scott fired him over coffee at Juice, the company she was running. He asked why nobody told him, and said he thought they all cared about him. Juice failed not long after, and Scott calls it the low point of her career. She draws a straight line from missing guidance to a team that gets poor results.

Most managers believe the risk in feedback is coming across as harsh. Scott argues the expensive failure runs the other way. Caring about a person and declining to challenge their work has a name in the book: Ruinous Empathy. She blames it for the vast majority of management mistakes in her career.

The Framework

The book runs on two axes. Care Personally means caring about each person as a whole human being rather than about their ability to do a job. Scott dates hers to Moscow on July 4, 1992, recruiting ten diamond cutters. Money moved nobody. What moved them was whether she got them and their families out if Russia collapsed. Caring personally is not about memorizing birthdays.

Challenge Directly means telling people when the work is not good enough, when it is, and when they do not get the role. Her rough test: if nobody is ever mad at you, you are not challenging the team enough. The harder half is inviting them to challenge you.

• Radical Candor combines both axes, with praise that is specific and criticism that arrives even after a win.
• Obnoxious Aggression challenges with no care shown, and Scott ranks it second best, because at least people know where they stand.
• Manipulative Insincerity fails both axes and pushes emotional buttons for personal gain, which makes it the worst quadrant.
• Ruinous Empathy cares personally and refuses to challenge, and decent managers default to it.

Two rules keep the grid from becoming a personality test. Guidance gets measured at the listener's ear, not the speaker's mouth. Directness in Tel Aviv reads as respect, and Tokyo needed reframing as politely persistent. The names describe guidance and never people, because labels stop improvement. Scott leaves three unimportant things unsaid each day.

Key Ideas

Correction Works Best Inside Two Minutes

Scott presented to Google's chief executive and founders, and it went well. Sheryl Sandberg walked her back and opened with specific praise, naming three or four things. Scott kept fishing for the criticism. Sandberg told her she said um a lot and asked whether she knew.

Scott waved it off with a shooing gesture twice, which forced the escalation. Sandberg said that saying um so much made her sound stupid. Google then paid for a speech coach, and three sessions produced noticeable improvement.

Scott lists why it worked. The correction came immediately, the praise was sincere with no sandwiching between bogus positives, and Sandberg escalated only after being ignored. The words landed on how Scott sounded rather than on who she was. Help came attached, and all of it fit inside two minutes.

Ruinous Empathy Is Where Decent Managers Live

The Russian version of the quadrant is a man who has to amputate his dog's tail. He loves the dog too much, so he cuts off an inch a day.

Three forces put managers there. People avoid tension at work. Most of us grew up with advice to say nothing when nothing nice comes to mind. Fear of the jerk label then pushes people into Ruinous Empathy and Manipulative Insincerity, both worse for colleagues than Obnoxious Aggression.

Scott tests the grid on a colleague whose fly is down. Telling him in private is Radical Candor. Announcing it for laughs is obnoxious and still beats silence, because he can fix it. Sparing his embarrassment costs him ten more witnesses.

The cost compounds twice. Reports hear good job, never learn where they stand, and stall or get fired. Bosses stop asking for criticism, so nothing looks wrong until someone quits.

Solicit Criticism Before You Give Any

Ask for criticism first, then give it. Doing so shows you know you are often wrong, teaches you what only your reports see, and improves your own delivery. One rule protects the practice: if a person is bold enough to criticize you, do not critique the criticism.

Her go-to question, borrowed from Fred Kofman, asks what the boss can do or stop doing to make working with her easier. Most people answer that everything is fine. Do not rescue them. Count to six before saying anything else. If six does not work, ask again.

Then reward what comes back. Scott handed her team a blue rubber band and let them snap it every time she interrupted. Toyota painted a red square on the assembly line floor. New employees stood in it after their first week and named at least three things to fix.

Tally praise against criticism, and treat all praise and no criticism as the warning. Past about sixty reports, waiting for a private moment means you never hear it, so let your most comfortable critic ask at an all-hands.

Guidance Is a Two-Minute Habit

Scott calls guidance the atomic building block of management. Be humble, using Situation, Behavior, Impact from the Center for Creative Leadership: the situation you saw, the behavior, and the impact it had. Be helpful, and show rather than tell, because retreating to abstractions is its own form of Ruinous Empathy.

Say it immediately, in the two or three minutes between meetings. Guidance has a short half-life, and unspoken criticism detonates later. With five reports, praising each three times a week and criticizing each once a week takes at most sixty minutes a week. The enabler is the calendar: twenty-five and fifty minute meetings with hard stops, instead of thirty and sixty.

Do not personalize. Say that the work is wrong rather than that the person is wrong. Scott names the fundamental attribution error, coined by Lee Ross, as the trap underneath. She calls the phrase do not take it personally worse than useless.

She rejects fixed praise-to-criticism ratios and the feedback sandwich, which Ben Horowitz gave a cruder name. A target ratio pushes people into saying things that are insincere. Her instruction inverts the usual worry: spend less effort on phrasing criticism and more on giving praise. Most managers overcorrect through Obnoxious Aggression before they settle in Radical Candor.

Growth Trajectory Replaces Potential

A leader at Apple told Scott that a team needs two kinds of excellent people, and was explicit about the balance. Rock stars are solid as a rock, in their groove, and uninterested in your job. Superstars sit on a steep growth trajectory and go crazy doing the same job a year later. Scott admits she undervalued the rock stars at Google.

The vocabulary changed during an argument with Scott Forstall over McKinsey's nine-box performance-potential matrix, built for GE to decide which businesses to fund. Scott objected that no human being has low potential. They replaced potential with growth, which turns the question into what trajectory each person wants right now.

The grid plots past performance against future growth trajectory, and the bottom right is as good a place as the top right. Steep growth is not a promotion and not a management track, and plenty of individual contributors stay steep for a whole career. Trajectories also move.

Rock stars get recognized and rewarded and not promoted, because promotion puts them in the job they do not want. The rewards are bonuses, raises, top performance ratings, go-to expert status, teaching offered as an honor, and tenure awards. Apple framed those at the five, ten, fifteen, and twenty year marks.

Superstars get new opportunities, mentors outside the team, and an instruction to teach their job to someone else. Google made that structural. People nominate themselves for promotion, a committee decides, the manager does not sit on it, and no boss can block a transfer.

The conversation that reads a trajectory is Russ Laraway's: three forty-five-minute sessions per report across three to six weeks, inside the normal one-on-one slot. The first asks for the life story starting with kindergarten, because values show up in the changes people made. The second asks for three to five dreams, including the crazy one, and turns the skill gaps into a curriculum. The third becomes an eighteen-month plan naming who does what by when.

Decisions Belong to the People Closest to the Facts

Results come from a wheel with seven steps: listen, clarify, debate, decide, persuade, execute, learn. Skipping a step wastes time later, and letting one drag turns collaboration into a tax.

Listening has two legitimate styles. One student of Scott's spent ten minutes of every one-on-one silent and expressionless, and heard the things he did not want to hear. Loud listening puts a strong opinion on the table, held weakly, and demands an answer. Scott let an ideas team assign her up to three action items a week. One member's programmable keypad raised the global team's efficiency by 133 percent.

Clarify by editing rather than authoring. Debate is a rock tumbler, and the friction polishes both the ideas and the people. Scott makes people argue the other side, warns them in advance so they listen harder, and fixes a date for when the debate ends.

The boss is usually not the decider. James March's garbage can decision making puts whoever happens to be at the table in charge instead of whoever holds the facts. Scott calls that a large driver of organizational mediocrity. The decider gets facts rather than recommendations, because people attach their egos to recommendations.

Persuasion needs the listener's emotion, the speaker's credibility, and the logic. Steve Jobs opened the 2003 iTunes for Windows announcement with a line about hell freezing over. Windows held over 90 percent of the market and the Mac under 5 percent.

Practical Applications

Put fifty minutes a week per direct report on the calendar, and give the agenda to them. One-on-ones exist to listen and clarify, and saved-up criticism does not belong in them. Ten reports drops the slot to twenty-five minutes a week, and twenty reports to twenty-five minutes every other week. Scott limits herself to five direct reports.

Run the staff meeting on a fixed clock. Twenty minutes review the key metrics. Fifteen minutes are study hall, where everyone writes three to five things their team did and reads everyone else's. Thirty minutes identify the coming week's decisions and debates. The meeting names those decisions and does not make them.

Spin the named debates and decisions into their own meetings with named owners. A big debate meeting lowers tension, because everyone knows no decision comes out of it. Its product is a summary of the facts and a recommendation. The big decision meeting follows, and its decisions are final.

Block think time and defend it. One chief executive Scott cites held two hours every day and declined a meeting with a president rather than move it. All-hands meetings start paying at one hundred people or more. The whole program costs about ten hours a week, five of which are one-on-ones most managers already hold.

Who This Is For

Managers of five to fifty people get the most from it, above all first-time managers told to give feedback and never told the words. Founders past the first ten hires get the meeting architecture and the firing rules.

Skip the relationships chapter, which is the softest in the book. Its one durable idea is the platinum rule: find what makes the other person comfortable and do that.

The framework is easy to misuse. Two seconds of care plus a hard truth reads as a license for cruelty. The care-personally axis is the half people skip, because it is slow and cannot be performed in a meeting. Scott forbids using the quadrant names as labels for people, and the labels are the part that spread. The evidence is one manager's experience at two large technology companies, reported without an audit.

The Decision

In your next one-on-one, ask what you can do or stop doing to make working with you easier. Then stop talking. Count to six while the silence sits there. If the answer is that everything is fine, ask again and wait through the second silence.

A real criticism means the channel is open. Reward it in public, and change one thing people can see. Nothing after two asks means you are the last person in the room who knows what is wrong. What comes back in that silence measures every piece of guidance you gave before it.