Career Frameworks

A career framework is the map that gives the rest of performance management a destination. Without one, "growth" is a vibe — people work hard with no idea what they're working toward, reviews judge against an invisible bar, and promotions feel arbitrary and political. With one, growth conversations point somewhere concrete, leveling decisions can be defended, and people can largely self-navigate.

The framework's job is to answer one question well: "What does the next level look like, specifically, and how do I get there?" Everything else is detail.

What a Career Ladder Is

A career ladder is a set of levels, each defined by the scope, complexity, and autonomy of the work a person handles — not by years served or lines of code. A good ladder describes behavior at each level concretely enough that two managers reading it would place the same person at the same level.

Typical engineering levels (names vary wildly; the shape is consistent):

The crucial property climbing the ladder: the work shifts from individual output to multiplied output. A junior is measured by what they ship; a staff engineer is measured by what the org ships differently because of them. Promotions stall most often when someone keeps doing more of the previous level's work harder, instead of changing what kind of work they do.

Defining the Levels Well

A framework is only as useful as its specificity. Good level definitions share traits:

• Behavioral, not credential-based. "Designs systems that other teams build on" beats "5+ years experience." Years are an input; the framework should describe outputs.
• Observable. A manager should be able to point at real work that demonstrates the level. If you can't cite examples, the definition is too abstract.
• Multi-dimensional. Most frameworks track several axes — technical skill, scope of impact, autonomy/ownership, communication, mentorship/influence. People are spiky across these, and that's expected.
• A center of mass, not a checklist. A level describes the typical altitude of someone's work, not a list of boxes all of which must be ticked. Treating it as a gate sequence ("you can't be senior until you've checked every senior box") produces unfair, mechanical decisions.

IC vs Management: Two Tracks

A critical structural decision: past a certain level, the ladder must fork into a dual track — individual contributor (IC) and management — that are genuinely equal in level, comp, and respect.

                      ┌── Principal Engineer (IC)
            Senior ───┤
                      └── Engineering Manager ──── Director ...

Why this matters:

• Management is a different job, not a promotion. Forcing your best engineers into management to advance loses a great engineer and often gains a mediocre manager. The two require different skills, energy, and satisfaction.
• The tracks must be peers. If "Staff Engineer" is quietly seen as less than "Manager" at the same level, the dual track is a fiction and people will feel pushed toward management anyway. Equal comp and equal voice are the test.
• The fork should be reversible. Someone who tries management and finds it isn't for them should be able to return to IC without it being a demotion or a stigma. Frame it as a lateral move between tracks.

When someone is deciding between the tracks, help them choose on the basis of what energizes them, not status. The honest question: do you get more satisfaction from solving hard technical problems yourself, or from making a group of people more effective? Neither answer is better.

Leveling

Leveling is placing a person at the right rung — at hire, and over time. Done badly it's the source of most comp inequity and resentment.

• Level the work, then the person. Anchor on the behavior in the definitions and the evidence, not on charisma, tenure, or how hard they negotiated at offer time.
• Down-level mistakes are expensive. Hiring someone a level below where they actually operate breeds resentment and attrition; over-leveling sets someone up to fail visibly. Both are corrected through calibration.
• Calibrate across managers. The same forces that distort reviews distort leveling — one manager's "senior" is another's "mid." Cross-manager calibration is the fix.
• No surprises at level boundaries. Someone should know whether they're "solidly mid," "approaching senior," or "operating at senior, not yet promoted" — never guessing.

Promotion Process and Packets

Promotion should be a recognition of a level already being demonstrated, not a reward you hope someone grows into.

The core principle: promote on sustained evidence of operating at the next level, not potential. "They could probably do it" is how you set people up to fail. "They have been consistently doing it for two quarters" is a safe, defensible promotion.

The promotion packet

The packet is the written case that someone is operating at the next level. A good one contains:

• A summary mapping the person's work to the next level's definitions, axis by axis.
• Specific evidence — projects, decisions, and outcomes that demonstrate next-level behavior, not just hard work at the current level.
• Peer and stakeholder input corroborating the impact, especially the multiplier effects (mentoring, influence, raising the bar).
• The manager's assessment of where they're solid and where they're still growing into the level.

Writing the packet well is part of the manager's job — and a vague packet is often a sign the person isn't actually ready, or that the manager hasn't given them the work to demonstrate readiness.

Common promotion dysfunctions

• Promoting for tenure. "They've been here three years" is not evidence of next-level scope.
• Promoting to retain. Using a level bump to stop someone leaving, when they're not operating at that level. It pollutes the ladder and others notice.
• The squeaky wheel. People who lobby get promoted; quiet high-performers stagnate. The manager's job is to advocate for the people who don't advocate for themselves.
• Surprise denials. Someone goes up for promotion and is denied with reasons they'd never heard. A failure of the manager's feedback and growth conversations.

Growth Conversations

The framework is dead paper until it's used in conversation. Growth conversations are where you and the report connect their day-to-day work to the map.

A simple recurring structure (often in a monthly/quarterly career-focused 1:1):

1. Where do you want to go? IC or management? What kind of work energizes you? Don't assume "up" — some people are happy and excellent at their current level.
2. Where are you now, honestly? Locate them against the framework, with specifics, including the spiky bits.
3. What's the gap to the next level? Name the specific next-level behaviors not yet consistently demonstrated.
4. What work would close it? This is the actionable part — the delegation of stretch work that lets them demonstrate the next level. Growth happens through the work, not the conversation about it.
5. What support do you need from me? Mentoring, exposure, air cover, sponsorship.

The most powerful move a manager makes here is sponsorship — actively handing someone the stretch project and the visibility that lets them demonstrate the next level, then advocating for them when the decision is made. Mentoring tells someone what to do; sponsorship gives them the chance to do it and makes sure the right people see it.

Anti-Patterns

• No framework at all. Growth is vibes, promotions are politics, leveling is negotiation skill.
• The framework as a checklist. Treating spiky humans as boxes to tick. Levels are a center of mass.
• Up as the only direction. Implying everyone must climb. A great senior engineer happy at their level is a gift, not a stagnation problem.
• Forcing engineers into management. The only path up runs through managing people. Loses an engineer, gains a reluctant manager.
• Promoting on potential. Rewarding hope instead of demonstrated behavior. Sets people up to fail in public.
• The invisible bar. A framework that exists but is never used in conversation, so people still don't know what next level means.

The Bar To Hold

A framework specific enough that two managers would level the same person the same way, with genuinely-equal IC and management tracks, where promotion follows sustained evidence of next-level work rather than tenure or potential, and where every report knows where they stand and what specifically would move them up.

The framework is the map; delegation is how people actually travel it; 1:1s are where you read it together.