§ CAPABILITY

Engineering Career Ladders

Levels with observable criteria, a management track and an individual contributor track that genuinely reach the same seniority, and a mapping of your existing people onto it.

Who
Led by the founder, hands-on for the duration.
Request a call How the work runs

What you're seeing

Two engineers doing comparable work are paid differently and neither knows why.
Usually means Every offer was negotiated from scratch. The inequity was not intended, it is invisible until it is discovered, and it tends to be discovered by several people at once.
A promotion conversation has no shared answer to what the next level requires.
Usually means There is no framework, so promotion is decided by advocacy. The people promoted are the ones with managers who argue well, which is not the same population as the ones who earned it.
The only way to earn more is to become a manager.
Usually means There is no genuine individual contributor track. The predictable result is strong engineers becoming reluctant managers, which loses you an engineer and produces a manager who did not want the job.
Titles were handed out to close offers.
Usually means Levelling is being used as negotiation currency. It works once per hire and the accumulated inconsistency surfaces the moment a framework is introduced.

Levels exist to make decisions comparable

Without them, every offer is negotiated from scratch and every promotion is decided by advocacy.

Neither is obviously wrong at the time. Each individual decision is defensible, made by reasonable people with the information available. The problem is the aggregate: two engineers doing comparable work on different money, one promoted because their manager argues well and another not because theirs does not, and no way to explain either outcome when it is questioned.

That aggregate stays invisible for about a year and then becomes visible all at once, usually because two people compared notes. At that point it costs considerably more to fix than it would have cost to prevent.

Observable, or it is a preference

“Demonstrates strong ownership” is an adjective, and adjectives get applied as preferences.

Two managers reading the same criterion about the same engineer will reach different conclusions, and both will believe they applied the framework correctly. That is not a failure of diligence; there is nothing in the sentence to converge on.

The three axes that can be written observably are scope — how large a piece of work this person can be handed — autonomy — how much direction they need and at what granularity — and influence — how far beyond their own output their effect reaches. Each of those can be described in terms that two managers reading the same evidence score the same way, and that convergence is the whole reason the framework exists.

The parallel track has to be real

Where the senior individual contributor levels pay less than their management equivalents, or carry fewer decision rights, engineers read that accurately.

They then take the management path in order to progress, which costs you a strong engineer and produces a manager who did not want the job and is unlikely to be good at it. Both outcomes are expensive and both are entirely predictable from the compensation table.

A genuine track means equal bands and equal standing at each level. That is a real commitment and some leadership teams are not ready to make it. If so, the honest position is a single track, clearly stated — which is better than a decorative branch nobody believes in.

Map before you publish

The moment a ladder is published, everyone applies it to themselves privately, and the anomalies are discovered by the people affected by them.

Mapping first — placing everyone, recording the reasoning, and surfacing the mismatches — means leadership finds the under-levelled and the over-titled before the announcement, and can plan the corrections and their budget. It is the uncomfortable part of this work and it is the part that determines whether the framework is trusted.

Where it sits

This capability sits under Engineering Hiring, where it is upstream of almost everything else: you cannot make consistent offers without knowing what level you are hiring at.

Two neighbours depend on it. Performance Review Design has nothing to anchor ratings to without a ladder, so reviews assess effort and likeability instead. And Team Topology & Org Design is the other half of retention — a boundary that gives people a clear scope, with no progression path attached to it, produces attrition rather than ownership.

How the work runs

  1. Decide the shape

    How many levels, where the senior threshold sits, and where the tracks separate. More levels means more promotion events and more calibration cost.

  2. Write observable criteria

    Scope of work, autonomy, and influence, in terms two managers would score the same. Criteria that read as adjectives get applied as preferences.

  3. Build the parallel tracks

    An IC track reaching the same levels and compensation as management. Where it does not, the ladder is a management track with a decorative branch and people read it correctly.

  4. Map the existing team

    Everyone placed, with the reasoning recorded, and anomalies surfaced before publication rather than discovered by an employee.

What arrives

  • A levelling framework with observable criteria per level
  • Parallel IC and management tracks reaching equivalent seniority
  • Salary bands aligned to levels
  • A mapping of the current team with the anomalies identified

What it costs your team

Two workshops with leadership, plus a mapping review per manager.

Next step

Tell us where this is breaking and we'll say what it takes to fix.

Request a call

How we decide

  • The existing team is mapped before the ladder is published

    Costs Mapping surfaces people who are at the wrong level, which is an uncomfortable and sometimes expensive discovery.

    A ladder published without a mapping is a document everyone immediately applies to themselves, privately, and the anomalies are then discovered by the affected employees rather than by leadership. Finding them first means the corrections and their budget can be planned. Finding them second means responding to them individually, under pressure, in a worse order.

  • Criteria are observable, never adjectival

    Costs Writing them properly takes two workshops and produces disagreement between managers.

    Criteria that read as adjectives — 'demonstrates strong ownership' — get applied as preferences, and two managers will score the same person differently while both believe they applied the framework. Scope of work, degree of autonomy and radius of influence can be described in terms two managers converge on, and that convergence is the entire point of having levels.

  • The IC track reaches the same levels and the same money as the management track

    Costs It commits the company to senior IC compensation that some leadership teams are not ready for.

    Where the top IC levels pay less or carry fewer decision rights, the track is correctly read as a consolation prize and nobody chooses it. You then lose strong engineers into management, which costs you the engineer and produces a manager who did not want the role. A decorative parallel track is worse than an honest single one.

  • Five or six levels, not more

    Costs Fewer levels mean each spans more, and some people sit inside one for years.

    More levels means more promotion events, more calibration cost, and a treadmill where progression becomes an annual expectation rather than a change in scope. Fewer, wider levels are easier to calibrate and honest about the fact that growing from mid-level to senior takes years rather than quarters.

Frequently Asked Questions

A framework describing the levels an engineer can occupy, what is expected at each, and what distinguishes one from the next — usually in terms of scope, autonomy and influence. It exists so that hiring, promotion and compensation decisions are made against a shared standard rather than case by case.
Around fifteen to twenty engineers, or earlier if promotion conversations have started and there is no shared answer to what the next level requires. Below that, the founder or the technical lead can hold the standard in their head; past it, they cannot, and pretending otherwise produces inconsistency.
Usually five or six through to staff. Fewer means each level spans so much that it stops being meaningful. More means constant promotion cycles, heavy calibration overhead, and an expectation that progression happens annually regardless of whether scope changed.
Only where the top IC levels carry the same compensation and the same decision rights as their management equivalents. Where they do not, people read the track accurately as a consolation prize and take the management path anyway. A decorative parallel track does more damage than not claiming to have one.
They surface during mapping, which is the awkward part and the reason to map before publishing. Some are under-levelled, which costs money to correct. Some are over-titled, which is harder. Both need a plan and a budget agreed before the framework is announced rather than after.

Sources

Page reviewed