A senior who announces a departure already holds an offer from a competitor. The counteroffer settles money and leaves everything else standing: next month he sits on a legacy module with no roadmap. He builds his next offer out of the weeks you give him here.
You lose more seniors to stagnation than to salary
Money shows up in every exit conversation because the competitor already put a number on the table. Twelve months later the engineer who kept the same scope accepts it. He spent two quarters fixing bugs and reviewing pull requests, and none of his decisions survived in the system.
Autonomy belongs on the same list. The senior picks the tool, sets the approach, and kills a weak solution in the meeting without asking three times. Once every choice routes through a committee, he delivers approved scope and stops offering opinions. Two months of that and he updates his own profile.
The work that stays on his calendar
A problem with no answer on file. Platform migration, a rebuild of the payment system, an integration with a 2009 ERP. Known-result work suits anyone; you hired this senior for reasoning.
A decision that leaves a mark. He wants to watch latency drop after the call he defended, infrastructure cost fall the next month, uptime rise the week of the deploy. Without a number attached to the decision, his work reads as opinion in the priority meeting.
A peer at his level. One senior alone is a single point of failure with nobody to argue against; he got a call wrong for a year and the mistake survived. Three seniors in the same room turn disagreement into a design review, and that review is where he decides to stay.
The retention routine you run
- Rotation every two quarters. Nobody stays past two quarters in maintenance. Move the senior between platform and product work; new scope restores the habit of deciding.
- A budget for conferences and courses. Registration, flights, and two days out beat a new perk. Fund open source contributions in working hours and the engineer stays close to the market.
- An open technical roadmap. A senior who knows where the platform goes over the next two quarters folds his own work into the plan and stops guessing priority.
- Recognition with a number. "The change you shipped cut infrastructure cost by 30%" beats any all-hands applause. Name the figure, name the person, repeat it in front of the team.
Will Larson describes the engineer who widens his own reach in Staff Engineer (O'Reilly, 2021). The more decisions he touches, the less he shops around. Google's Project Aristotle, published on re:Work, placed impact on one's own work among the five dynamics that separate effective teams.
You lose the senior when his work turns into predictable maintenance for two quarters. Rotation, an outside budget, an open roadmap, and recognition with a number cost less than a replacement search, and the replacement spends months learning why the payment table is shaped the way it is.
