Few things damage trust in a growing company faster than a promotion nobody can explain. One person is promoted after a year, another waits three; one manager pushes hard for their team, another never asks. Nobody needs to be acting in bad faith for the result to look unfair — it only takes each manager doing it their own way.
This article describes a promotion process for companies of 20 to 200 people without a large HR team: what the promotion criteria should be, who proposes and who decides, what a written case contains, and how to tell people the outcome. It assumes you have, or are building, a career ladder with levels and written expectations. Without one, there is nothing to be fair against.
What "fair" means in practice
A fair promotion process does not mean everyone is promoted at the same pace. It means five things:
- The criteria are known in advance. Everyone can read what the next level asks for before anyone is assessed against it.
- The same criteria apply to everyone. A promotion in support is judged the same way as one in engineering, against that department's own expectations.
- The decision is not made by one person. The manager who knows the person best proposes; someone with a wider view decides.
- The reasoning is written down. Six months later, anyone involved can explain why the decision went the way it did.
- The person is treated with care. A pending promotion is not half-announced, and a "no" comes with a clear path forward.
Most unfair processes fail on the first or the fourth: criteria in managers' heads, or decisions nobody recorded.
Set the promotion criteria first
The promotion criteria should be the expectations of the target level, and nothing else. If your career ladder says that a level 3 support specialist "is the go-to person for at least one complex product area and writes internal notes others rely on", then that is what a promotion to level 3 must show.
A practical standard used by many teams: promote people who are already working at the next level, consistently, across most of its competencies. Not people who might grow into it, and not people who did one exceptional project. "Consistently" usually means several months and more than one piece of work.
Equally important is what does not count:
- Time in role. Useful context, never a criterion.
- A single result. One great quarter in sales or one heroic release is a data point, not a level.
- Retention pressure. A promotion to match an outside offer skips the criteria everyone else is held to. If you need to pay someone more to keep them, that is a pay decision — make it as one.
- Visibility. People who present well are noticed more; written expectations make quieter work count too.
Where a department measures output with numbers — quota in sales, tickets resolved in support — use them as evidence for the relevant competency, not as the criterion itself. A skills matrix that sets the level each role requires for every skill makes the criteria concrete: the case is then measured against that role's row, and nothing else.
Who proposes, who decides
Separate the two roles.
The direct manager proposes. They see the work every week, they hold the career conversations, and they will have to explain the decision either way. A skip-level manager or a peer can suggest that someone is ready, but the proposal comes from the person's own manager.
Someone above them decides. The manager's own manager, a head of department, or in a small company a founder. Their question is not whether every detail is right but whether this case meets the same bar as the others they have approved. Seeing cases from several teams, they are the natural check on a generous or a strict manager.
Employees can raise it; managers write it. Encourage people to say "I think I'm working at the next level — here is why" in a career conversation. Then the manager either agrees and writes the case, or explains what is missing. Self-nomination as a formal route tends to reward the confident rather than the ready.
In a company of 25 where the CEO manages everyone, the separation is thin — but writing the case down and waiting a week before deciding still gets most of the benefit.
Write the promotion case
A written case is the single biggest step towards fairness. It forces the manager to argue from the criteria, and it gives the decider something to compare. Keep it to one or two pages:
- Target role and date. Which role, on which track, from when.
- The expectations, one by one. For each competency of the target level: does the person meet it, and what is the example? Two or three concrete examples per competency beat a paragraph of praise.
- The history. What recent check-ins said — ratings over time, gaps that have closed. A pattern across two or three check-ins is stronger than a single one.
- Other voices. What colleagues have said in feedback, especially people outside the manager's team.
- The gaps. Where the person does not yet fully meet the level, and why the manager thinks it should not block the promotion. A case with no gaps at all is usually a case that did not look hard enough.
A short example, for a support specialist moving from level 2 to level 3:
Product and technical knowledge (level 3: go-to person for a complex area). Since March, Maria has handled every escalated billing-integration ticket; the team's internal notes on billing are hers, and two colleagues now resolve those tickets alone using them.
Support operations (level 3: proposes and runs changes for the team). Rewrote the refund macro set; repeat contacts on refunds fell noticeably in the following two months.
Gap: has not yet trained a new hire end to end. A new specialist starts in November and Maria will onboard them.
Make the decision
The decider has three outcomes, not two:
- Approve, with an effective date.
- Ask for more — the case may be right but the examples are thin. Send it back with a specific question.
- Reject — the person is not yet working at the target level. Say which expectations are missing.
To keep standards consistent across teams, the decider should keep the last few approved cases for each level in mind and compare. Once or twice a year, put all the promotions of the period side by side — by level and department — and ask whether any of them would look out of place next to the others. That review is a conversation among managers; it does not need special tooling.
LadderKit's promotions workflow follows this shape: the direct manager writes the case for any published role, starting from the person's profile or a completed check-in, and anyone above that manager in the reporting line can approve it, reject it or send it back asking for more. The person never sees the request about themselves, and once it is approved the new role applies on the effective date.
Tell people the outcome
How you communicate the decision matters as much as the decision.
Do not half-announce. A person told "I've put you forward" who is then rejected has been handed a disappointment they did not need. Most companies are better off keeping a pending case between the managers involved until it is decided. The criteria are public; individual cases are not.
Make the promotion real on a date. Agree on an effective date and change the role, the title and — if your ladder has pay ranges — the pay together, on that date.
Make "not yet" useful. When a case is rejected, or when someone raised it and the manager does not agree, the conversation should name the specific expectations that are missing, with examples, and end with two or three concrete actions with due dates. An individual development plan built from those gaps turns a "no" into a plan. Agree when you will look at it again — and then do.
A worked example: a 45-person company
A software company of 45 people: 20 in engineering, 8 in sales, 6 in support, the rest in product, design and operations. Before the process, promotions happened when someone asked loudly enough or received an outside offer. Two engineers were promoted after threatening to leave; a support specialist who had quietly become the team's expert was not.
What they changed. They published a ladder for each department. They agreed that the direct manager writes the case and that the head of department decides — the CEO decides for the heads themselves. They set one rule: no promotion to match an outside offer.
The first round. Over six months, five cases are written. Three are approved. One — an account executive with an excellent quarter — is sent back with a question: the case showed results but nothing on the next level's expectations for deal management and closing. The manager adds two examples and it is approved a month later. One — a product manager — is rejected; the conversation names two missing expectations from the product manager ladder, and two growth actions follow.
The review. At the half-year mark, the CEO and the department heads spend an hour comparing the five cases. They agree on a shorter common format: the engineering cases were twice as long as the rest.
The result. The support specialist who had been passed over is promoted in the first round, on a case that pointed at three level 3 expectations with examples.
Common mistakes
- Criteria written after the candidate is chosen. The case then describes the person, not the level.
- Matching outside offers with promotions. Everyone who waited their turn notices.
- One decider per team. If each manager decides for their own team, you have as many standards as managers.
- "Not yet" without specifics. People leave over vague refusals more than over clear ones.
- No record. If nobody can say in a year why a promotion was approved, the next one will be argued from memory.
Where to go from here
The process only works if the levels underneath it are clear: if they are not, start with how to build a career ladder. If you run separate individual-contributor and management tracks, read how a move between them relates to promotion in the dual career ladder. And because good cases are built from months of conversations, regular check-ins are the best preparation for your next promotion decision.