1.3 Roles, career ladders, and growth

Overview and motivation

A career ladder is the explicit map of roles, levels, and expectations that tells engineers what “good” looks like at each stage and how to grow. In a small company, growth stays informal: the founder knows everyone, promotions are ad hoc, and titles barely matter. At scale, that informality turns into a liability. Without a shared, written framework, promotion decisions get political, expectations swing wildly from manager to manager, and engineers cannot tell whether they are making progress. A clear ladder turns the vague question “am I doing well?” into a concrete one you can actually answer, and it turns promotion from a favour a manager grants into a defensible, evidence-based decision.

For large teams, career frameworks solve three problems at once. They bring fairness and consistency across hundreds of people and dozens of managers who would otherwise each apply their own standards. They open dual career paths, so senior engineers who want to deepen their technical impact are not forced into management to advance. And they make hiring, levelling, and compensation coherent, so “senior engineer” means roughly the same thing everywhere in the organisation. The alternative, levelling by vibes, breeds resentment, pay inequity, and the loss of exactly the people you most want to keep.

Enterprises and government organisations have their own reasons to formalise this. Regulated environments demand defensible, documented decisions, promotion and pay equity included. Government pay systems and job classifications are often rigid, which makes mapping real engineering competency onto a formal grade especially important to get right. And because these organisations keep people for long careers, the ladder is not just a promotion tool. It is your main instrument for long-term workforce development, mentorship, and succession planning.

Key principles

Provide dual tracks: individual contributors and managers advance in parallel, with equal prestige and pay at senior levels.
Levels describe scope and impact, not tenure or activity.
Competencies must be written, observable, and applied consistently through calibration.
Growth is a shared responsibility: the engineer drives it, the manager enables it, the org resources it.
Sponsorship, not just mentorship, moves careers; ensure it is distributed equitably.
Hiring should be structured and evidence-based to reduce bias and improve prediction.
Titles are a means to clarity and fairness, not a reward in themselves.

Recommendations

Define parallel IC and management tracks

Establish two tracks of equal standing. The management track advances through people leadership, team health, and organizational outcomes. The individual-contributor (IC) track advances through technical depth, architectural influence, and multiplying the effectiveness of others. The crucial part: make senior levels equal across tracks in pay and prestige, so a principal engineer and a director are peers. This heads off the classic failure where the only way to earn more is to stop doing the technical work you are great at. That failure loses your best engineers and hands you reluctant, ineffective managers.

Articulate staff-plus archetypes

Beyond senior, IC roles branch into recognisable archetypes rather than a single “more senior engineer.” Common ones include the Tech Lead (guiding a team’s execution), the Architect (owning technical direction across a domain), the Solver (parachuting into the hardest problems), and the Right Hand (extending a senior leader’s reach across an organisation). Naming these helps engineers and managers agree on what impact looks like at these levels, because it really is different from just writing more code. Be clear that scope, influence, and judgment define these levels, not lines of code.

Build a competency matrix and calibrate it

Write a competency matrix that spells out, for each level, the expected behaviours across dimensions like technical skill, scope of impact, autonomy, communication, leadership, and business awareness. Keep the descriptions concrete and observable (“influences technical decisions across multiple teams”) rather than vague (“is very senior”). Then calibrate: managers review proposed ratings and promotions together, against the matrix and real evidence, to normalise standards across teams. Calibration is what stops an easy manager’s “senior” from meaning less than a strict manager’s, and it is where fairness actually gets enforced.

Invest in mentorship, sponsorship, and apprenticeship

Keep mentorship and sponsorship separate in your mind. Mentorship is advice and guidance. Sponsorship is actively advocating for someone, putting your own credibility behind them for opportunities and promotions. Mentorship helps people improve; sponsorship is what actually moves careers forward. So track it and distribute it deliberately, or it will flow only to the people who resemble your existing leaders. Use structured on-call apprenticeship to build operational maturity safely: new engineers shadow experienced responders, then take primary with a backup, then mentor the next cohort. That turns on-call from a dreaded burden into a deliberate growth ladder.

Structure hiring to reduce bias and improve signal

Unstructured interviews mostly measure how much a candidate resembles the interviewer. Replace them with structured ones: the same job-relevant questions and rubrics for every candidate, work-sample exercises that look like the real job, and independent scoring before any discussion so no one anchors on the first opinion. Train your interviewers, use diverse panels, and keep the evaluation of evidence separate from the final decision. Define the level and competencies you are hiring for before you interview, so you assess against a target instead of a gut feeling. And document your decisions, both for fairness and for legal defensibility.

Trade-offs: pros and cons

Decision	Pros	Cons
Dual IC/management tracks	Retains technical talent; better managers by choice	More levels to maintain; risk of unclear IC impact
Detailed competency matrix	Consistency, fairness, clear growth path	Can become rigid box-ticking; costly to maintain
Formal calibration	Normalises standards; reduces bias	Time-consuming; can feel bureaucratic
Structured hiring	Better predictions; less bias; defensible	Slower to design; less “gut feel” flexibility

The core tension is structure versus flexibility. Too little structure gives you bias, inconsistency, and politics. Too much gives you box-ticking, where people optimise for the rubric instead of real impact and the framework hardens even as the work changes underneath it. The way out is to treat frameworks as descriptive guides for judgment, applied by calibrated humans, not rigid checklists scored by rote. The ladder should inform your promotion conversations, not replace them.

Questions to discuss with your team

Is your individual-contributor track equal to management in pay and power, or equal only on paper? The whole point of dual tracks is to stop losing your best engineers to management they never wanted, and that promise fails silently when a principal engineer earns less, or carries less real influence, than a director at the same level. Check the evidence: compare actual compensation bands across the two tracks at senior levels, count how many people have reached staff-plus versus how many directors exist, and notice who is actually in the room for consequential decisions. In government settings, rigid civil-service classifications make this harder, so the mapping from engineering competency to official grade has to be deliberate and documented. If the IC track pays less or stays quieter in decisions, engineers will read the real message and route back into management. The answer tells you whether to fix comp bands, decision rights, or both before you promote your next principal engineer.
How will you detect and correct inequitable sponsorship, given that it flows invisibly by default? Mentorship is advice and sponsorship is advocacy, and sponsorship is what actually moves careers, which means unequal sponsorship quietly reproduces your current leadership’s makeup across hundreds of people. You cannot manage what you do not measure, so bring data: who gets the visible stretch projects, who gets named in promotion calibration, and how those opportunities distribute across demographic groups and across teams. For a large or regulated organisation, this is both a fairness obligation and a legal-defensibility one, because pay-equity and promotion-equity claims turn on exactly this pattern. The trade-off is that tracking sponsorship feels intrusive and adds overhead, while the cost of not tracking is a narrowing talent pool and regretted attrition. If the distribution is skewed, assign sponsorship deliberately rather than leaving it to who resembles the existing leaders.
When the nature of the work changes, how will you keep the competency matrix from hardening into a checklist people game? A matrix is meant to be descriptive guidance for calibrated human judgment, and its failure mode is ossifying into a rubric people optimise for while the real work shifts underneath it. Ask what “good” looks like now versus when the matrix was written: as tooling, automation, and architecture evolve, behaviours like “writes more code” matter less and judgment, scope, and multiplying others matter more, which is exactly what the staff-plus archetypes describe. Bring evidence of gaming: promotions that satisfied the rubric but did not correspond to real impact, or engineers doing high-value work the matrix does not describe. In enterprise and government settings the pressure to treat the matrix as a rigid box-ticking form runs strong because it feels defensible, and that rigidity is precisely what breeds resentment. Revisit the matrix on a cadence, keep it a guide for promotion conversations rather than a replacement for them, and let calibration committees weigh real evidence over rote scoring.
How do you know your calibration process actually normalises standards, rather than just ratifying whatever each manager already decided? Calibration is where fairness is supposed to get enforced, yet it degrades easily into theatre, where managers rubber-stamp one another’s ratings to avoid conflict and an easy manager’s “senior” quietly outranks a strict manager’s. For a large organisation spanning dozens of managers, this drift is the difference between a defensible framework and a political one, so the committee has to challenge evidence against the matrix rather than nod it through. Bring the data: the distribution of ratings and promotion rates by manager and team, how many proposed levels actually changed during calibration, and concrete cases where the committee overturned a recommendation. In enterprise and government settings this record doubles as your audit trail, since promotion and pay decisions may have to withstand oversight or a legal challenge, and if nothing ever changes in the room, the process is calibrating nothing and the fix may be training, tighter evidence standards, or committee composition.
Who is accountable for resourcing growth, and is that responsibility funded or merely assigned? Growth is a shared responsibility across the engineer, the manager, and the organisation, and the organizational share is the one that silently goes unfunded the moment delivery pressure rises. Decide deliberately how much time managers and senior engineers may spend on mentorship, structured on-call apprenticeship, and sponsorship, then protect it, because “grow your people” written on a manager’s goals with no time attached is a promise nobody can keep. Bring evidence: the actual hours spent on development versus delivery, whether apprenticeship rotations happen or quietly slip, and how growth investment tracks against regretted attrition and internal mobility. In government and other long-tenure organisations this investment is your main instrument for workforce development and succession planning, so under-funding it now surfaces years later as a leadership and skills gap you cannot hire your way out of in time.
Do your interviews predict on-the-job performance, or mostly measure resemblance to the people already here? Unstructured interviews reliably reward candidates who look and sound like the interviewer, which narrows your talent pool while feeling like sharp judgment. Moving to structured interviews, the same job-relevant questions and rubrics, work-sample exercises, and independent scoring before any discussion, trades some gut-feel flexibility and design effort for better prediction and less bias. Bring the evidence: whether you define the target level and competencies before interviewing, how interview scores correlate with later performance and promotion, and how hiring outcomes distribute across demographic groups. For enterprise and government hiring, documented and structured decisions are also a legal-defensibility requirement, because a challenged rejection or a fairness audit turns on whether every candidate was assessed against the same job-relevant standard.

Sector lens

Startup. A full ladder is overhead a tiny team cannot afford, so write a single page: three or four levels, a few observable expectations for each, and a note that pay follows scope. Skip calibration committees, competency matrices, and staff-plus archetypes until headcount forces them on you. But write something down the moment two engineers ask what “senior” means, because vibes-based levelling breeds resentment fast as the team doubles.

Small business. With no dedicated people function and a tight budget, borrow rather than build: adapt a public engineering ladder, such as the progression.fyi collection, instead of authoring one from scratch. Keep it lightweight and spend where it pays back, on structured, consistent interviews that avoid a costly mis-hire, and on a growth path clear enough to keep the handful of engineers you could not easily replace. Treat the ladder as a retention tool, not paperwork.

Enterprise. Scale makes consistency the hard problem: hundreds of engineers and dozens of managers who would each apply their own standard. Invest in a detailed competency matrix, formal calibration across teams, dual IC and management tracks with genuine pay equity at senior levels, and staff-plus archetypes so technical impact has somewhere to grow. Track promotion equity, regretted attrition, and internal mobility as portfolio metrics, and keep the matrix a guide for judgment rather than a box-ticking form that hardens as the work changes.

Government. Rigid civil-service classifications and strict fairness rules shape every choice. Map your engineering competency matrix onto the official job grades deliberately and document the mapping, so engineers get a transparent, defensible path within an inflexible pay system. Rebuild hiring around structured, job-relevant work samples with independent scoring and diverse panels, and keep an audit trail of every promotion and hiring decision, because oversight and legal defensibility depend on it.

Examples

Startup. At a fifteen-person startup, titles are casual and growth happens through direct founder feedback, which works fine until two engineers ask, in the same month, what “senior” would mean and what it would pay. Rather than import a corporate ladder, the founders write a single page: three levels, a handful of observable expectations for each, and a note that pay follows scope. It takes an afternoon and heads off the resentment that vague, vibes-based levelling breeds as the team doubles. The lightweight sketch is enough for now, and it gives everyone something concrete to grow into.

Enterprise. A global software company kept losing its best senior engineers because the only path to higher pay ran through management, and many of them had no wish to manage. It introduced a staff-plus IC track with named archetypes and a competency matrix, explicitly pegging principal engineer to director and distinguished engineer to VP in compensation. Calibration committees now review IC promotions with the same rigour as management ones. Regretted attrition among senior engineers dropped, and technical decision quality rose, because the organisation’s deepest experts stayed in technical roles instead of drifting into management they never wanted.

Government. A government technology unit needed to modernise hiring while bound by rigid civil-service classifications and strict fairness requirements. It mapped its internal engineering competency matrix onto the official job grades, then rebuilt interviews around structured, job-relevant work samples with independent scoring and diverse panels, documenting every decision for auditability. This made hiring both fairer and more predictive while satisfying oversight requirements, and the documented competency mapping gave employees a transparent, defensible path to advance within an otherwise inflexible pay system.

Business case: motivations, ROI, and TCO

The biggest hidden cost a career framework addresses is regretted attrition. Losing a strong engineer costs a substantial multiple of their salary once you count recruiting, onboarding, lost productivity during ramp-up, and the institutional knowledge that walks out the door. Much of this attrition is avoidable. It comes from unclear growth paths, promotions that feel unfair, and the absence of a technical career track. A well-run ladder attacks those causes directly. Structured hiring does the same for the enormous cost of mis-hires and the slower, subtler cost of biased hiring that narrows your talent pool.

The adoption cost is moderate: mostly a one-time build plus ongoing calibration overhead. You write the matrix, train managers and interviewers, and run calibration cycles. That is weeks of effort and a few hours per review cycle, not a large capital outlay. The cost of not adopting is paid continuously, in attrition, pay-inequity claims, slow and biased hiring, and the quiet demoralisation of people who cannot see a future. To persuade leadership, connect the framework to metrics they already track: regretted attrition rate, time-to-fill and quality-of-hire, promotion equity across demographic groups, and internal mobility. Frame the ladder as workforce infrastructure with a clear payback, not as HR paperwork.

Anti-patterns and pitfalls

Management as the only way up: forces technical talent out of technical work.
Levelling by tenure: rewards time served rather than scope and impact.
Vague competencies: “is senior” language that invites bias and inconsistency.
Skipping calibration: lets standards drift wildly between managers.
Mentorship without sponsorship: people get advice but never advocacy or opportunity.
Unstructured interviews: measure cultural similarity, not job capability.
Ladder as checklist: engineers game the rubric instead of pursuing real impact.
Sponsorship hoarding: advocacy flows only to those who resemble current leaders.

Maturity model

Level 1, Initiate. No written ladder; promotions and pay are ad hoc and personality-driven, and “senior” means whatever a given manager decides.
Level 2, Develop. A basic ladder and some competency descriptions exist, but they are applied inconsistently across teams; there is no calibration, and hiring stays largely unstructured.
Level 3, Standardise. Dual IC and management tracks, a documented competency matrix, calibration, and structured hiring are standard practice and enforced org-wide, with senior levels pegged for pay equity across tracks.
Level 4, Manage. The framework is measured against baselines: regretted attrition, promotion and pay equity across demographic groups and across managers, time-to-fill and quality-of-hire, and internal mobility are tracked, and calibration overturns ratings on evidence rather than ratifying them.
Level 5, Orchestrate. Career development is integrated with workforce planning and succession; the matrix and archetypes are revised continuously as the work changes; sponsorship and apprenticeship are deliberately assigned and rebalanced, and equity gaps trigger corrective action rather than reporting alone.

Ideas for discussion

Can our best engineers advance without becoming managers, with equal pay and status?
Do our level definitions describe observable impact, or just seniority language?
How consistent are promotion standards across our managers, and how do we know?
Who receives sponsorship here, and does it flow equitably?
Do our interviews predict job performance, or mostly measure similarity to us?
Is our ladder guiding real growth conversations, or has it become a checklist to game?

Key takeaways

Provide equal, parallel IC and management tracks so technical talent can advance technically.
Define levels by scope and impact, described in concrete, observable competencies.
Calibrate ratings and promotions across managers to enforce fairness.
Sponsorship, not just mentorship, advances careers; distribute it equitably.
Structure hiring to improve prediction and reduce bias, and document decisions.
Treat the ladder as descriptive guidance for human judgment, not a rigid checklist.

References and further reading

Will Larson, “Staff Engineer: Leadership Beyond the Management Track” (staff-plus archetypes)
Camille Fournier, “The Manager’s Path”
Tanya Reilly, “The Staff Engineer’s Path”
Lara Hogan, “Resilient Management”
Iris Bohnet, “What Works: Gender Equality by Design” (structured, bias-reducing hiring)
Google re:Work guidance on structured interviewing and calibration
Jonny Burch and the progression.fyi collection of public engineering ladders
Kim Scott, “Radical Candour” (feedback and growth)