The resignation feels sudden. It almost never is.
By the time a strong performer tells you they are leaving, the decision is typically three to six months old. What you are receiving is not the decision. It is the administrative confirmation of a decision made in a period during which, from the outside, nothing appeared to change.
That period is the only part of this that is actionable, and it is the part most organizations never see.
What the numbers say it costs
Replacement cost is routinely underestimated because companies count the recruiter and forget everything else. Research by Boushey and Glynn at the Center for American Progress puts the median at 21% of annual salary, rising to 213% for executive positions.
The visible portion (advertising, agency fees, interview time) is the smaller half. The rest is the vacancy period, the ramp-up during which the replacement produces below capacity, the institutional knowledge that walked out undocumented, and the effect on the people who stayed and are now doing two jobs while watching a good colleague leave.
Gallup's engagement meta-analysis, across more than 183,000 business units, found that units in the top quartile of engagement show 51% lower turnover than those in the bottom quartile. The difference between a team people want to stay in and one they do not is not marginal. It is half your attrition.
Why the exit interview tells you nothing
Exit interviews are conducted by the organization the person is leaving, in a room where the person has nothing to gain from candour and a reference to protect.
So you get the safe answer. Compensation. Growth opportunity. A new challenge. All plausible, all partially true, and all describing the offer that closed the decision rather than the conditions that opened it.
The real sequence usually runs like this. Something happened: a project killed without explanation, a promotion given to someone visibly less capable, a boundary crossed once too often. The person raised it, or did not feel able to. Nothing changed. They stopped raising things. Six weeks later, they answered a recruiter they would previously have ignored.
Compensation is what makes leaving possible. It is rarely what makes leaving desirable.
The distributed version: nothing to notice
In an office, the pre-resignation period has a physical signature. Someone books unusual mid-week afternoons off. They are dressed differently on a Tuesday. They stop volunteering. Managers register these without articulating them, and the good ones act on the feeling.
Distributed teams delete all of it. There is no dress, no corridor, no calendar you can read. Availability is the only visible signal, and availability stays perfect right up to the last day, because a professional finishing their tenure delivers cleanly.
There is a second factor that is specific to remote work and rarely stated: your best people have the largest market. A senior engineer in a distributed company is not competing with the local employment market. They are visible to, and reachable by, every company in their timezone band. The cost of leaving you has dropped, while the cost of retaining them has not.
The three conditions that actually hold people
Across our diagnoses, the teams with low regretted attrition tend to share three things. None of them is money.
They are seen. Their contribution is visible to people who matter, and named. Recognition here is not a programme with a quarterly award. It is a manager who can describe, specifically, what this person did that mattered.
They have somewhere to put their judgement. They are allowed to decide how their work gets done. Autonomy over method is the single most protective factor we observe, and it is cheap.
They belong to something. They can say who "we" are and mean it. This is the one distance erodes fastest and the one companies invest in least, because it does not fit in a policy.
Where a diagnosis changes the picture
The reason retention work usually fails is that it starts from a guess. Leadership assumes the issue is compensation, benchmarks salaries, raises the band, and loses the same people twelve months later, because the problem was recognition, or a manager, or the absence of any reason to stay beyond the paycheque.
Our diagnosis measures those conditions separately, and in the distributed version compares what leadership believes with what the team reports. The gap is usually where the leaving happens.
The people you are about to lose are not going to tell you. But the pattern that is producing them is measurable now, while they are still here.
LUNITA