Career Frameworks for Scale-ups: When and How to Build One

Career frameworks attract two opposite failures. Some companies build one at thirty people, complete with eight levels and a competency matrix, and spend a year maintaining a structure the organisation is not big enough to need. Others reach two hundred people with no levels at all, at which point every salary is a historical accident and the framework has to be retrofitted around decisions nobody would make again.

The useful question is not "are we big enough" but "are we making the decisions a framework exists to support".

The four triggers

1. You cannot answer "what would it take to get to the next level". If a good performer asks this and the honest answer is "it depends", you are already paying for the absence.
2. Pay differences within the same work are hard to explain. Not wrong necessarily — just unexplainable, which has the same effect the first time two people compare notes.
3. Promotion is the only recognition mechanism you have. Titles inflate because there is nothing else to give, and you end up with four "Heads of" in a team of twenty.
4. Hiring is setting internal precedent. Every offer benchmarked externally without an internal structure creates a comparison you will have to defend later.

Two of these together is usually the point to build. In practice that lands somewhere between sixty and a hundred and twenty people for most companies, but the range is wide and driven by hiring pace rather than size.

Build fewer levels than you think

The most common design error is too many levels. Each one has to be distinguishable in practice, defensible in a promotion case, and fundable. A level that cannot be described without using the word "slightly" should not exist.

Add the parallel technical track only when you have at least two people who genuinely belong on it. Publishing a technical track nobody occupies is worse than not having one: it makes an implicit promise the company then fails to keep.

Write levels in scope, not in years

Level definitions written in years of experience are unusable, because they measure the wrong thing and can never be argued with. Four dimensions do the work:

• Scope — the size of the problem this person is trusted with. A task, a project, a system, a function.
• Autonomy — how much direction they need. Told what and how, told what, asked what should be done, deciding what matters.
• Complexity — how ambiguous and how interdependent the work is.
• Impact — who is affected when they do it well or badly. Their own work, their team, adjacent teams, the company.

Write each level as a paragraph a person could read and recognise themselves in, then test it: give three managers the same anonymised profile and ask them to place it. If they disagree, the definitions are not distinguishing what you think they distinguish.

Connect to pay deliberately

A career framework without pay ranges creates expectations the company has not costed. Ranges without levels create pay decisions nobody can defend. Build both, or accept that you have built half a system.

The minimum viable version: a range per level, a stated market positioning, a rule for where new hires enter the range, and a rule for how someone moves within it as distinct from between levels. Then a rule for who approves an out-of-range offer and how it is recorded.

The last point is where frameworks quietly die. Exceptions are not the problem; unrecorded exceptions are. Three undocumented out-of-range offers and the framework is no longer describing how the company pays.

Placing existing people

This is the hardest part and the part most likely to be rushed. Expect to find people who are paid above the range for the level they are actually at, and people who are below it. Both are normal and both need a plan before anything is communicated.

1. Place people on level first, using the definitions, without looking at salary.
2. Then overlay pay and identify the outliers in both directions.
3. Decide the correction approach for those below range, with a timetable.
4. Decide the approach for those above range — usually held rather than reduced, with slower progression.
5. Brief every manager on their own team before anything is published.

A framework is judged entirely by the first three promotion decisions made under it. Get those right and it becomes the standard. Get them wrong and it becomes a document.

What to publish

Level definitions and promotion criteria should always be visible to everyone: their entire purpose is to answer a question people currently cannot get answered. Salary ranges are a separate decision, and a legitimate one either way — but publish nothing until managers can explain the framework in their own words. The most common failure is announcing transparency and leaving managers to field questions they were never briefed on.

If you are weighing this now, the careers and compensation work is the same project, and doing them together is materially cheaper than doing them a year apart.