Employee Retention

For a long time, people leaving was treated as a string of personal circumstances. Someone gave notice, their manager felt bad about it, the rest of the team absorbed the work, and HR posted the role again. Taken one at a time, the reasons really were all different. One person went back to school, one had a spouse transferred to another city, one got a call from a larger company. Nothing about any single departure suggested there was anything to manage.

Collect six months of them and the picture changes. In a 45-person company, eleven people leaving in a year with seven of them from the same team is not eleven personal circumstances. The cost of refilling those seats shows up at the same moment. Replacing one person and getting them up to speed routinely costs more than half of that person's annual salary. The hours spent writing the posting and running interviews, the work the remaining team splits while the seat sits empty, and the months the new hire needs to reach full output all land inside that figure.

The real gap was that nobody owned the work. Bringing people in had an owner, a headcount target, and a budget line. Keeping the people already there had none of those. So companies moved only after hearing that someone was leaving, which is where the familiar scene comes from — a manager saying "hang in there a bit longer, let me see what I can do about your salary" in a hurried one-on-one. By then the other person is usually holding a written offer, which narrows the company's options down to money alone. An exit interview will eventually surface the real reason, but that conversation happens after the decision is already made.

Retention runs as a cycle that turns before anyone gives notice, usually once a quarter. There are four steps.

1. Find where the leaks are, in numbers. Break turnover rate down by team, by level, and by tenure. If the company sits at 15 percent and one team sits at 40, that team is the target. If departures cluster in the first year, the problem lives in hiring or in early onboarding rather than in anything a raise would fix.
2. Ask why. A one-on-one, the recurring private conversation between a manager and a team member, catches individual signals. An engagement survey, which asks everyone the same questions about satisfaction and commitment, catches patterns at team level. The first tells you about one person's situation; the second tells you what several people are hitting at once.
3. Pick one or two things to change. Common candidates are re-fitting the pay band — the salary range written down for each role and level — writing an individual development plan that says what the next step is and what has to be filled in to get there, and adjusting when and where someone works. Don't do all three at once. Change everything simultaneously and you learn nothing about which one worked.
4. Read the same number two cycles later. A team that went from 40 percent to 22 gives you something to keep doing. A team that didn't move means step 1 pointed at the wrong place, and it's worth looking again.

The order gets inverted constantly. A company skips the asking and starts at step 3, adding company-wide perks nobody requested, and the money moves turnover barely at all.

The primary owner is the direct manager, because what usually decides whether someone stays is the daily work and the people sitting next to them. HR collects the numbers, runs the surveys, and maintains the salary ranges. The founder or executive owns the levers that need authority — budget, and any change to the bands themselves. A manager who can see the problem but can't move a number ends up offering sympathy instead of a fix, and people can tell the difference.

The timing is planned in advance, because the moments when people make up their minds cluster in a few places.

1. Three months and one year in. If expectations and reality came apart, these are the two points where it surfaces.
2. Right after a performance review result is delivered. Someone who receives a lower rating than they expected tends to decide that week.
3. Right after a promotion announcement. Look at the people who expected one and didn't get it, not at the people who did.
4. Right after a large project ships. Several months of pushing hard leaves people inclined either to rest or to move.

"We're scoping this quarter's retention work to the engineering team. Their turnover is double the company average."

The problem has been narrowed to one team rather than treated as a company-wide issue.

"You don't keep someone by responding to a resignation letter. You keep them with what came up in last month's one-on-one."

A push back on acting only after notice is given.

"Salary alone won't hold him. He has no idea what he gets to work on next."

The diagnosis is a missing growth path, not a number.

At Pagedeck this came up the week after review results went out. Grace was breaking turnover down by team, saw that engineering was running far above everyone else, and sat down with Theo. What Theo brought up wasn't money. It was that his work had been identical for two straight quarters.

Grace didn't answer with a raise. Instead the two of them picked one area Theo would own for the next quarter and wrote down what he'd need to show there to reach the next level. The compensation conversation came after that. Had the order been reversed, Theo would have taken the raise and left six months later for the same reason he was already unhappy.

The most common mistake is reading retention as another word for raising salaries. People watch departures happen over a higher offer, so the association forms easily. But pool the stated reasons and money is always one item among several, while the relationship with a manager and the absence of a visible next step come up more often. Block a departure with cash alone and the same person starts looking again within six months, for the same reason as before.

Treating zero turnover as the goal is the other one. A lower number looks unambiguously better, so it gets written down as a target. But an organization nobody leaves has no seats opening either, and the way it works calcifies. Low performers stay put along with everyone else. What matters is not the size of the number but who is inside it. Fifteen percent where three of the ten departures were people the team depended on is worse than 25 percent made up of people who were barely holding their roles.