He Hits His Numbers

Somewhere in a building you have worked in, a manager is being told about a problem.

She is not a monster. She is tired. She is two people short, the quarter closes in five weeks, and the person sitting across from her has just spent eleven minutes describing something that she can tell is real and cannot quite picture herself writing down. A pattern. A tone. Something about meetings.

While she listens, a part of her mind she would not admit to is running a separate calculation, and it goes like this.

If I open this, it takes four months. Two of those months are mine. If it goes the way these go, I lose him. He's the one who cleared the backlog in March. I'd be down three, not two, in the worst quarter of the year, and I'd be explaining to my own boss why I broke a functioning team over something I can't put on paper.

She says she's glad he came to her. She says she'll keep an eye on it.

She means both. She will also do nothing, and she will not experience herself as having decided that.

This is the part of the machine that runs on ordinary people making locally sensible choices. Nobody in this scene is a villain, and that is precisely why it works.

Most of us carry an unexamined belief about how organizations work, and it survives contact with reality for a remarkably long time.

The belief is that being good at your job and being decent to people are two separate accounts, and that an institution audits both. That a person can be brilliant and still be removed over how they treat people, because the treatment is its own line item, assessed on its own terms.

Almost no organization is built that way. One of those accounts produces a number that appears in a report every quarter. The other produces nothing that appears anywhere at all.

In 2015, two researchers ran the numbers on this in a way nobody had before.

Michael Housman and Dylan Minor, in a Harvard Business School working paper using data from a talent-management firm, looked at 58,542 workers across 11 firms and 184 sub-firms, all in frontline service positions, and tracked them from hiring assessment through employment to exit. They defined a "toxic worker" narrowly and verifiably: someone eventually terminated for an egregious violation of company policy. Harassment. Falsifying documents. Fraud. Violence. About 4.5% of the workforce ended up in that category.

Then they asked the question everyone assumes they know the answer to.

Toxic workers were more productive. Not less. Faster than the average employee, measurably so, though the quality of what they produced was lower.

Sit with that, because it dismantles the comfortable story. The reason organizations keep these people is not that leadership is blind, or spineless, or in denial. It is that the manager in the first scene is looking at something real. He did clear the backlog in March. The output is not a fiction she has been tricked into believing. It's on the dashboard.

Which means every conversation that begins "how do they not see it" has the premise wrong. They see it. They are weighing it against something they also see, and one of those two things arrives in a report.

Here is where the arithmetic turns.

Housman and Minor calculated what a toxic worker actually costs, and to keep the estimate defensible they counted almost nothing. Not litigation. Not lost customers. Not the meetings, the investigations, the quiet erosion of everyone's willingness to speak. Just turnover: the cost of replacing the people who leave because of this person.

On that basis alone, avoiding one toxic worker was worth about $12,489.

For comparison, they ran a figure on the thing every organization actually spends its energy chasing. Hiring a superstar, a top one-percent performer, was worth about $5,303.

Those two numbers measure different quantities, and it is worth being precise about that rather than letting the contrast do work it hasn't earned: one is avoided turnover cost, the other is added productivity value. They are not the same currency. But the order of magnitude is the point. The cheapest, most conservative accounting of what one such person costs still comes out larger than the most-chased prize in corporate hiring.

There is one more finding, and it is the one that should frighten anyone running a team. Workers with greater exposure to a toxic colleague showed a 46% increased hazard of becoming toxic themselves. It spreads. What looks like one difficult person is, on a long enough timeline, a transmission event.

So if the math is that lopsided, why does the manager in the first scene get it wrong?

Because she isn't seeing that math. Nobody is.

The benefit of keeping him is concentrated, immediate, and attributed. It has his name on it. It appears in her numbers, this quarter, where her own performance is assessed.

The cost is diffuse, delayed, and anonymous. It shows up as three resignations spread across fourteen months, each one filed under a different reason. Personal circumstances. Better offer. Relocation. It shows up in a team that has quietly stopped raising problems early, which registers as nothing at all, because you cannot measure the meeting where somebody decided not to speak.

No organization on earth generates a report that reads: this quarter we lost four people, two years of institutional knowledge, and the willingness of a department to flag defects, because of one employee who is excellent at his job. The accounting doesn't exist. So the cost is real, it is enormous, and it is invisible in exactly the place where the decision gets made.

The manager is not being irrational. She is optimizing correctly against the only ledger she can see.

Now the caveat, and it is a real one.

The Housman and Minor study defines toxicity by termination-worthy conduct. Fraud. Harassment. The things that eventually produce a firing. That is a much narrower and more extreme category than what most people are actually enduring at work.

You might think this weakens the argument. It does the opposite.

If a worker whose conduct was severe enough to eventually get them fired was nonetheless kept around long enough to be measured, then the tolerance threshold sits somewhere above that line. And everything documented in Part 2, the exclusion, the deniable tone, the story with no author, sits far below it. Those behaviors will never generate a terminating event. They produce no violation to cite. They simply run, indefinitely, underneath the floor of anything an organization has been built to notice.

The study cannot tell us how large that lower region is. Nobody can, which is a problem this series returns to at the end. What it can tell us is where the threshold sits, and the threshold sits above conduct serious enough to eventually justify a firing.

Which brings us to the sentence I would put in front of anyone who has ever been on the receiving end of this and concluded that nobody believed them.

You were not being disbelieved. You were being priced.

Those feel identical from the inside and they are not remotely the same. Disbelief is a factual dispute, and factual disputes can be resolved with evidence. Pricing is a judgment that the facts are probably accurate and are outweighed. No amount of documentation touches it, which is why people who finally build the airtight case are so often devastated to find that it changes nothing. They brought proof to a negotiation they didn't know they were in, about a value they weren't told they'd been assigned.

That is a harder thing to hear than "they didn't believe me." It is also, in most cases, the truth, and I would rather hand you the accurate version.

Notice, finally, that we have a vocabulary ready for this person, and that it is entirely admiring.

Demanding. Direct. Doesn't suffer fools. High standards. Hard to work for, but you learn a lot. Every one of those is a way of describing conduct while declining to price it, and every one is available to a manager who needs to narrate a decision they have already made.

We have no comparable vocabulary for what he costs. There is no phrase that does for the quiet departure of four good people what "doesn't suffer fools" does for the man who caused it. The language is asymmetric in exactly the direction the accounting is, and for exactly the same reason.

So here is the question, and it is aimed at anyone who has ever had to make the call.

Think of the difficult person on your team, the one everyone works around. You have almost certainly done the arithmetic already, even if you never said it out loud.

What number did you put on the other person?

Sources

• Housman, M., & Minor, D., Toxic Workers, Harvard Business School Working Paper 16-057. Sample of 58,542 workers across 11 firms and 184 sub-firms in frontline service positions; toxicity defined as involuntary termination for egregious policy violation, approximately 4.5% of the sample. Reported figures: approximately $12,489 in avoided turnover cost per toxic worker avoided, against approximately $5,303 in value from hiring a top 1% performer; 46% increased hazard of becoming toxic with greater exposure to a toxic colleague.
• Summary and discussion via Kellogg Insight, Northwestern University, and Harvard Business Review.

The opening scene is a composite. Identifying details throughout this series have been removed or changed.