I have spent the past decade in healthcare staffing, which means I have spent a decade watching people change jobs.

Staffing gives you an unusual view of turnover. You hear why people leave before their employer does, because they tell the recruiter the real reason and tell their manager something polite. And inside our own company, every recruiter or account manager who walks out takes relationships and pipeline with them, so we have had to get good at keeping people ourselves.

Today I am President and COO of Uniti Med, GQR Healthcare and Nebula. While we grew Uniti Med from a startup, the company was named one of Fortune’s Best Workplaces. Before staffing I spent ten years in restaurants and then running sales and marketing for a brewery, an industry where turnover is so normal people stop seeing it. What follows is what I have learned from both sides about how to reduce employee turnover in a way that holds.

What causes most employee turnover?

Most people do not leave a company. They leave a situation: a manager who does not know them, a job that turned out to be different from what was described, a schedule that does not work for their family, or a sense that they have stopped getting better and nobody has noticed. Money is usually part of the story they tell, and it is real, but it is rarely the whole story.

The useful way to think about pay is as a threshold. If you are well below market, people will leave and nothing else on this page will save you. Once you are in a fair range, the next few dollars from a competitor only win when something else is already wrong. I wrote about that in You Are Not Going to Win on Pay, and I wrote about the other side, what actually makes people stay, in What Actually Keeps People.

Before you change anything, put a number on what turnover is costing you. Our Employee Turnover Cost Calculator takes your headcount, turnover rate and average salary and shows the yearly cost and what you would save by bringing it down. Leaders tend to treat turnover as a cost of doing business until they see the number written down.

Why are managers the biggest lever on turnover?

Because people quit people, not companies. It is the title of one of my keynotes, People Quit People, Not Companies, and it is the single most reliable thing I know about retention. The same pay, benefits and building can produce a team that stays for years under one manager and a revolving door under another.

That is good news, because a manager’s habits can be taught. The managers who keep people do a few ordinary things every week: they hold a real one-on-one, they know what each person is working toward, they notice effort before results show up, and they protect people in public and correct them in private. The managers who lose people usually are not bad people. They were promoted for being good at the work and never shown how to lead the people doing it.

When we were small, the first leadership hire at Uniti Med was not another salesperson. It was a Director of Culture. I wrote about that decision in Forbes. It sent a signal that how people were led would be measured the same way the number was measured, and that is the real fix. If you only hold managers accountable for output, you will get output and turnover. Hold them accountable for retention on their team too.

How do you reduce turnover in the first 90 days?

Early turnover is the most expensive kind, because you paid to recruit and train someone and got almost nothing back. It is also the most preventable. People who leave in the first few months are usually telling you that the job did not match the description, or that nobody seemed ready for them when they arrived.

A strong first 90 days does not need a big budget. It needs a plan someone actually owns:

Before day one. Call or text the new hire the week before. Tell them where to park, who they will meet first and what the first day looks like. Silence between the offer and the start date is where second thoughts grow.

Day one. Their manager should be there, not on vacation or in back-to-back meetings. Equipment, logins and a schedule should be ready. A first day spent waiting for a laptop tells someone exactly how much they matter.

The first two weeks. Give them a peer, not just a trainer, someone they can ask the questions they are embarrassed to ask their boss. Check in at the end of week one and week two with one question: is this job what you expected it to be?

Days 30, 60 and 90. Short, scheduled conversations about what is working, what is confusing and what they want to learn next. By day 90 they should be able to describe one thing they are better at than when they started.

What is a stay conversation, and why does it work?

An exit interview asks someone why they left after it is too late to do anything about it. A stay conversation asks the people you want to keep what would make them leave, while you can still act. It is the cheapest retention tool there is and most organizations never use it.

Keep it simple and keep it human. A few questions are enough: What keeps you here? What would make you think about leaving? What is one thing about your job you would change if you could? What do you want to be doing a year from now? Then do something with at least one answer, and tell the person you did. A stay conversation with no follow-through is worse than none, because it proves that asking is just a ritual.

Does recognition really affect retention?

Yes, though almost nobody resigns over recognition directly. They resign after a slow pile-up of good work that nobody noticed, and then tell you it was about the money. Recognition is how people find out whether their effort is registering anywhere.

What works is specific, soon and from someone who knows the work. Not “great job this quarter,” but “the way you handled that upset family on Thursday kept them with us.” It costs nothing and takes about a minute. I built my keynote The Power of Small Moments around this idea, because culture is built in the ordinary interactions nobody thinks are important. For a list you can use, see Employee Recognition Ideas That Cost Almost Nothing.

How do growth paths keep people from leaving?

People leave when they stop growing. If the only way to get a raise, a new title or a new challenge is to take a job somewhere else, your best people will eventually do exactly that, usually the ones you most wanted to keep.

A growth path does not have to be a formal career ladder, although those help. It can be as simple as each person knowing the next skill they are working on and the next role they could grow into, and a manager who asks about it in every one-on-one. Our best leaders at Uniti Med mostly came from inside, the recruiters the rest of the desk already went to for help long before anyone gave them a title. Promoting from within tells everyone else that staying is a strategy, not a sacrifice. I wrote more about this in What If You Do Not Train Them and They Stay?

Can schedules reduce turnover?

In many industries, the schedule is the job. In healthcare, restaurants, manufacturing and retail, a person might love their team and still leave because their shifts change every week or they cannot get a weekend off for their kid’s game. From the staffing side, control over when you work is one of the main things contract and agency work offers people, and it is a big part of why so many make that move.

You do not need to solve every scheduling preference. You need predictability and some control. Post schedules further out. Let people trade shifts without three approvals. Protect the commitments people have told you matter to them. In healthcare in particular, I have written about this in Nurse Retention Strategies That Actually Work, because it is often the difference between a nurse who stays and one who becomes a traveler.

What should you stop doing?

Stop relying on exit interviews. They are the least reliable data you have, because people leaving rarely want to burn a bridge on the way out.

Stop building fences. Clawbacks, non-competes and making it painful to leave do not keep people engaged. They keep people resentful until the fence comes down. I wrote about this for Inc. in Stop Building Fences to Keep Your Employees.

Stop letting the best people carry the load. Reliable people get handed the extra shift, the difficult customer and the new hire to train, often with no thanks. Then they burn out and leave, and everyone is surprised.

Stop treating turnover as an HR metric. Turnover is a leadership metric. Put it on the same scorecard as revenue and quality, broken out by manager, and talk about it in the same meeting.

Where should a leader start?

Pick the team with the highest turnover and look at the manager before you look at anything else. Then run the first three stay conversations yourself, this month. You will hear the real reasons people are thinking about leaving, and you will walk out with two or three things you can fix before the next resignation letter lands. If you want help bringing this to your managers, the employee retention keynote and the leadership workshop are built for exactly that.