Conventional retention metrics measure who stayed. Headcount at the end of the year versus the start. Attrition rate. Whether the team is the same size in December as it was in January.
Those are useful for HR dashboards. They’re a poor proxy for whether you’re building something worth being part of.
What the metric misses
A team where everyone stays isn’t necessarily healthy. It might mean the team is good. It might also mean no one has better options, the work isn’t stretching anyone, or people are coasting in a comfortable enough environment to not bother leaving.
The failure mode I think about most isn’t people leaving. It’s two quieter ones: people leaving because the environment drove them out, and people staying but disengaged — present on the org chart, absent from the work that matters. Both are expensive, and neither shows up cleanly in a retention number.
What I’ve found more predictive: whether people who have worked with you before want to work with you again, and whether the people who left went somewhere they’re growing faster than they would have stayed.
Who follows you
When I moved to a new program, a new client, a new initiative, there was always a question of team composition. Some of it was assigned; some of it involved people choosing to join.
The engineers who asked to come along — who had options and picked this program — told me something. Not that I was particularly good at cheerleading or making promises, but that the environment we’d built before was worth repeating. That work was real, that ownership was real, that the failure modes were handled honestly.
That’s a signal worth tracking. Not formally — no one has a spreadsheet for “voluntary follows” — but noticeably. If the people whose judgment you respect, and who have options, keep choosing the programs you’re on, something in the environment is working.
If they stop, something has changed, and the number to watch is not the headcount.
Who outgrows you
The other signal is people leaving for something bigger than you could offer them.
I’ve had engineers who came in as strong contributors, grew into tech-lead roles, and then moved on — to other organisations, in other countries — to what was the right next thing for them. Some were positions we couldn’t offer internally. Some were opportunities that matched where they wanted to go, and staying would have been limiting.
I count those as successes. Not because losing someone strong doesn’t hurt — it does, and the knowledge transfer and handover take real time — but because the direction of the movement matters. Someone leaving to grow is different from someone leaving because they were pushed.
A healthy team produces this kind of attrition. The people who came in and were challenged, who got real ownership and worked through hard problems, will eventually build towards something beyond what the current environment can give them. If they never do, either the people weren’t growing or the roles weren’t stretching.
The failure mode worth preventing
The attrition worth preventing is the other kind: people leaving because the environment made it unreasonable to stay.
Unclear ownership, where no one knows who decides and every decision requires escalation. Work that doesn’t connect to outcomes anyone cares about. A culture where mistakes are hidden rather than examined. These drive out people who have options first — which means they reliably degrade the team’s capability before the numbers show it.
This is the part that’s actually in a leader’s control. The people who outgrow you — that’s a success. The people who leave because you couldn’t build an environment worth staying in — that’s the number to fix.
What this looks like in practice
As Acting Head of DevOps at Ollion since July 2026, I’ve been thinking about this more explicitly. Running a team of 16 engineers across different programs and client engagements, the signal I’m watching isn’t the headcount. It’s whether the people I’m responsible for are working on things that stretch them, whether ownership is clear enough that people feel the work is theirs, and whether the environment is honest enough that failures become learning rather than liability.
When those conditions hold, the retention takes care of itself — and so does the growth. The people who should stay will stay. The people who should move on to something bigger will do so having built something real while they were here.
The metric isn’t who’s on the roster at the end of the year. It’s the quality of the trajectory — for the people who stayed and the ones who left.