Retention & Career Growth

Key Dimensions

Dimension	Retention Signal	Attrition Signal
Growth	“I’m learning and being challenged”	“I’ve been doing the same thing for 18 months”
Impact	“My work matters and I can see it”	“I don’t know if anyone uses what I build”
Belonging	“I like my team and my manager”	“I feel isolated / I don’t trust my manager”
Compensation	“I’m paid fairly for my market value”	“I found out I’m 20% below market”
Autonomy	“I have ownership and decision-making power”	“Every decision goes through three approval layers”
Purpose	“The company’s mission resonates with me”	“I’m just shipping features for metrics”

What Actually Drives Attrition

The research:

Culture Amp’s analysis of 1,000+ companies found the top predictors of attrition:

1. Lack of career growth / learning (strongest predictor)
2. Relationship with direct manager
3. Lack of confidence in leadership
4. Compensation (below market)
5. Workload / burnout

Notice: compensation is #4, not #1. People don’t leave for money first — they leave for growth, then rationalize it with a salary bump. But compensation becomes the primary driver when someone discovers they’re significantly below market (15%+), because it signals disrespect.

The tenure curve:

Attrition risk is not uniform over time. The highest-risk periods:

• 3-6 months — “this isn’t what I expected” (onboarding failure)
• 18-24 months — “I’ve learned what I’m going to learn here” (growth plateau)
• After a major life event — new child, relocation, partner’s job change (context shift)
• After vesting cliff — stock/RSU vest creates a natural exit point
• After a reorg — new manager, new team, broken relationships

The manager effect:

Gallup’s research (12,000+ business units): managers account for 70% of variance in employee engagement. People don’t leave companies, they leave managers. This is true, but incomplete — people also leave because of companies (bad strategy, layoffs, toxic culture) even with great managers. Your management quality is the controllable variable.

Career Ladders — Designing the Growth Path

The dual-track ladder:

Every mature engineering org needs an IC (Individual Contributor) track and a management track, with roughly equivalent levels and compensation at each stage.

IC Track	Management Track	Scope	Compensation Band
Junior Engineer	—	Task-level	Band 1
Mid Engineer	—	Feature-level	Band 2
Senior Engineer	Engineering Manager	Team/project-level	Band 3
Staff Engineer	Senior EM	Multi-team/domain	Band 4
Principal Engineer	Director	Org-wide	Band 5
Distinguished Engineer	VP Engineering	Company-wide	Band 6

Common ladder mistakes:

The management-only path: If the only way to advance is management, your best ICs either become bad managers (Peter Principle) or leave. Spotify, Google, and Netflix all have staff/principal IC paths with director-level compensation.

Vague level definitions: “Demonstrates technical leadership” — what does that actually mean? Good ladders define behaviors with examples:

• Weak: “Provides technical guidance to others”
• Strong: “Identifies and drives resolution of cross-team technical debt; recent example: led the API versioning initiative across 4 teams, defining the standard and shepherding adoption over 2 quarters”

No staff+ path: Many mid-size companies have Senior as the terminal IC level. This creates a ceiling that drives senior attrition. You don’t need to call it “Staff Engineer” — but you need a level above Senior that recognizes broader impact without management.

Rigid gatekeeping: Promotion requires checking every box on a long list. Real growth is uneven — someone might be Staff-level in system design but Senior-level in mentorship. Evaluate the whole picture.

How top companies define levels:

Stripe: Publishes an internal “RFC culture” where Staff+ engineers are expected to write company-shaping RFCs. Level is defined by the scope and impact of decisions you drive, not by years of experience or number of reports.

Google: Uses a committee-based promotion process (the “promo committee”) that reviews a packet including peer feedback, project impact, and a manager’s justification. The committee is cross-functional, reducing manager bias. Criticism: it incentivizes “promo-driven development” — choosing projects for visibility over impact.

Amazon: Promotion to Senior and above requires a written document (the “promo doc”) that maps accomplishments to Leadership Principles. The bar is high and intentionally creates a bottleneck — Amazon has relatively few Principals compared to Google or Meta.

Individual Development Plans (IDPs)

An IDP is a written agreement between manager and report about the person’s growth focus for the next 6-12 months. Most IDPs are bad because they’re vague, manager-driven, or filed and forgotten.

What a good IDP contains:

Element	Good Example	Bad Example
Growth goal	“Move from strong Senior to early Staff by demonstrating cross-team technical leadership”	“Improve skills”
Specific skills	“System design for distributed data pipelines; written communication for architecture proposals”	“Be more technical”
Learning activities	“Lead the Q3 data platform RFC; take the Designing Data-Intensive Applications reading group; shadow the platform team’s architecture review for 4 weeks”	“Take some courses”
Stretch assignment	“Own the cross-team API standardization project (Staff-level scope)”	“Work on harder problems”
Success criteria	“By Q4: authored 2 architecture RFCs, positive feedback from partner team leads, promo packet started”	“Show improvement”
Manager support	“I will: introduce you to platform team leads, review your RFC drafts, sponsor your presentation at the engineering all-hands”	“Let me know if you need anything”

IDP cadence:

• Create: At the start of a growth cycle (after a review, or when someone joins your team)
• Check in: Monthly in 1:1s — “How’s the RFC going? Do you need more time carved out?”
• Revise: Quarterly — goals may shift as circumstances change
• Archive: When the growth goal is achieved or the person moves on

Stay Conversations — Proactive Retention

Stay conversations (also called “stay interviews”) are the opposite of exit interviews. You have them with people you want to keep, before they start looking.

The 5 stay conversation questions:

1. “What do you look forward to each day?” — reveals what they find meaningful
2. “What are you learning?” — if the answer is “nothing new,” they’re at risk
3. “Why do you stay?” — the answer tells you what to protect
4. “When was the last time you thought about leaving?” — normalizes the question; the answer reveals triggers
5. “What can I do to make your experience better?” — actionable, specific

When to have stay conversations:

• Quarterly with your top performers (they’re the ones being recruited)
• After a reorg — disruption shakes loose people who were previously stable
• After a peer leaves — attrition is contagious; departures trigger others to reflect
• At vesting cliffs — RSU vesting at 12/24/36 months creates natural decision points
• When you notice disengagement — quieter in meetings, fewer PRs, less initiative

The counter-offer trap:

When someone comes to you with an outside offer, you’ve already lost the retention battle. Counter-offers have a 50-80% failure rate — meaning half to most people who accept a counter-offer leave within 12 months anyway. Why? Because the underlying reasons they looked (growth, management, culture) don’t change because the salary went up.

The right approach: If someone you want to keep tells you they’re considering an offer, don’t counter immediately. Ask: “What’s driving this? If money weren’t a factor, would you still be considering it?” If it’s genuinely about compensation, match or beat the market proactively. If it’s about growth or fit, address those honestly — sometimes the right answer is “I think this move could be great for you.”

Compensation Philosophy

Comp as hygiene factor:

Herzberg’s Two-Factor Theory: compensation is a “hygiene factor” — when it’s inadequate, it causes dissatisfaction. But increasing it beyond adequacy doesn’t create motivation. The practical implication: pay people fairly (at or slightly above market), then invest your energy in growth, autonomy, and purpose.

The comp transparency spectrum:

Level	What’s Shared	Companies	Tradeoff
Full transparency	Everyone’s exact salary is public	Buffer, Gitlab	Eliminates comp grievances; constrains negotiation
Band transparency	Salary bands per level are published	Stripe, Google	People know if they’re fairly placed; managers can’t play favorites
Process transparency	How comp decisions are made is explained, but individual numbers aren’t shared	Most large companies	Reduces suspicion without full exposure
Opaque	No one knows anything	Old-school companies	Maximizes employer negotiating power; breeds resentment

Recommendation: Band transparency at minimum. When people discover (and they always discover) that peers are paid more for the same work, the trust damage is severe. Publishing bands eliminates the worst cases.

Equity refresh grants:

For companies using equity (RSUs, options), the 4-year cliff creates a predictable attrition pattern. Spotify and Google address this with annual “refresh grants” — additional equity grants each year that create overlapping vesting schedules. The effect: there’s always unvested equity creating golden handcuffs, and the refresh acknowledges continued contribution.

Internal Mobility — The Retention Safety Valve

When someone wants to leave your team for another team in the company, your instinct as a manager is to feel betrayed. Suppress that instinct. Internal mobility is a retention mechanism — the alternative is they leave the company entirely.

The Google approach:

Google allows engineers to transfer after 12 months in a role, with the receiving manager’s approval. The current manager can’t block a transfer (though they can request a transition period). This policy explicitly acknowledges that career growth sometimes requires a new context, and keeps talent in the company.

The anti-pattern:

Managers who block internal transfers — either through bureaucracy, guilt (“you’re leaving the team in a tough spot”), or political sabotage (badmouthing the person to the receiving manager). This is career sabotage and people remember it forever. Even if you keep them temporarily, they’ll leave the company at the first opportunity.

How to handle it well:

1. Thank them for telling you directly (rather than going around you)
2. Ask what’s driving the desire to move — is it growth, the work, a relationship issue?
3. If the move is right for them, actively help — make introductions, give a strong internal reference
4. Negotiate a reasonable transition period (4-8 weeks, not 6 months)
5. Use the departure to reflect on your team’s growth environment — what could you have offered?

The Engagement Stack

Think of retention as a stack — each layer must be solid before the one above matters:

+----------------------------+
|    Purpose & Meaning       |  <- "My work matters beyond this company"
+----------------------------+
|    Growth & Challenge      |  <- "I'm becoming better at my craft"
+----------------------------+
|    Belonging & Connection  |  <- "I like my team and manager"
+----------------------------+
|    Safety & Fairness       |  <- "I'm paid fairly, not at risk of surprise layoff"
+----------------------------+
|    Basic Needs             |  <- "Reasonable hours, decent tools, WFH flexibility"
+----------------------------+

You can’t compensate for a broken lower layer by investing in a higher one. A person who feels underpaid won’t be motivated by a company mission. A person who doesn’t trust their manager won’t be engaged by a stretch assignment. Fix the stack from the bottom up.

Anti-Patterns in Retention

Anti-Pattern	Why Managers Do It	Why It Fails
Retention through fear	“You won’t find anything better”	People resent it and leave spitefully
Counter-offer as strategy	Cheaper than recruiting	50-80% leave within 12 months anyway
Title inflation	Give a title bump instead of addressing real issues	Devalues titles, doesn’t fix growth gaps
The golden cage	Massive equity grant with long vest	People stay but disengage — “quiet quitting”
Ignoring top performers	“They’re fine, focus on the problems”	Top performers are the most portable
One-size-fits-all perks	Free lunch, ping pong, beer fridge	Different people value different things

References

Books

• The Manager’s Path — Camille Fournier (career ladders, growing ICs, the dual track)
• An Elegant Puzzle — Will Larson (systems thinking about career frameworks, org growth)
• First, Break All the Rules — Gallup/Buckingham (the Q12 engagement questions, strengths-based management)
• Drive — Daniel Pink (autonomy, mastery, purpose as motivation drivers)
• No Rules Rules — Reed Hastings (Netflix’s top-of-market comp philosophy)
• Staff Engineer — Will Larson (what Staff+ looks like, how to grow into it)

Research & Articles

• Gallup Q12 — the 12 questions that predict engagement and retention
• Culture Amp — “Employee Attrition Prediction” (top drivers of turnover)
• Herzberg’s Two-Factor Theory — hygiene factors vs. motivators
• “Brag Documents” — Julia Evans (practical self-advocacy for career growth)
• levels.fyi — the transparency tool that changed comp conversations in tech

Talks

• Patrick Lencioni — “The Truth About Employee Engagement” (the 3 signs of a miserable job)
• Will Larson — “Designing an Engineering Career Ladder” (StaffEng talks)
• Jessica Kerr — “The Origins of Opera and the Future of Programming” (on growth environments)