Salary bands for SMEs: building grades when you have never had any

A founder showed me two salaries. Same role, same team, joined fourteen months apart, one paid noticeably more. The only honest explanation was that the second candidate had negotiated harder in a month when the role had been open for a while. Both employees now knew.

This is the normal state of a business that has never had grades. It is not a failure of fairness — it is the arithmetic of hiring one person at a time against whatever the market was doing that month.

Start from what you pay, not from a survey

The instinct is to buy market data first. Resist it for now. Market data tells you what other businesses pay; it does not tell you what your own structure is, and without that you cannot tell whether you have a market problem or an internal consistency problem. In most SMEs it is the second.

1. 1

   List every role and what it is actually paid

   One line per person: role, department, years in the business, current fixed pay. Not a range — the actual figure. This table is the whole input.

2. 2

   Group roles by what they require, not by title

   Ignore designations, which in most SMEs are historic. Group by the level of judgement, autonomy and consequence the role carries. You will usually find four to six natural clusters in a business under two hundred people.

3. 3

   Name the grades

   Four to six, neutrally. Grade 1 through 6, or letters. Avoid names that imply seniority in a way that collides with existing titles, because that starts a separate argument you do not need yet.

4. 4

   Draw the band around what you already pay

   For each grade, look at what people in it actually earn. The band is roughly that spread, cleaned up — a minimum, a midpoint and a maximum. A common width is a maximum around one and a half times the minimum, which gives room for progression within a grade.

5. 5

   Find the outliers

   People paid above their band maximum and below their band minimum. Both matter, and the first is the harder problem.

6. 6

   Decide the correction plan

   Not a repricing. A plan, over cycles, described below.

Structure on one page

GRADE   TYPICAL ROLES              MIN      MID      MAX
-------------------------------------------------------------
G1      ______________________     ___      ___      ___
G2      ______________________     ___      ___      ___
G3      ______________________     ___      ___      ___
G4      ______________________     ___      ___      ___
G5      ______________________     ___      ___      ___

Band width: MAX is about 1.5x MIN, so someone can progress
within a grade for several years without a promotion.

Overlap: each grade's range overlaps the next by roughly a
third, so a strong performer in G2 can out-earn a new G3.
That overlap is deliberate and is what stops every pay
conversation becoming a promotion conversation.

POSITION IN RANGE tells you more than the number:
  below MIN   -> correction needed
  MIN-MID     -> developing in the role
  MID-MAX     -> experienced, performing
  above MAX   -> outlier, plan required

The two outlier problems

Below the minimum is the easy one: correct it, at the next cycle or immediately if the gap is large, and say why. People rarely object to being brought into a structure.

Above the maximum is the one businesses get wrong. Do not cut pay. The workable approach is to hold the salary and let increments go to the rest of the grade until the band catches up, and to say this to the person honestly rather than leaving them to notice a pattern of small increments. Handled badly, this is how you lose someone who was doing nothing wrong except being hired in a tight month.

What to publish

Publish the structure — that grades exist, roughly what sits in each, how progression within a band works, and when reviews happen. Do not publish individual salaries or the band figures themselves unless you are confident about every position in range.

The benefit of publishing the structure is that pay conversations change shape. "Why am I paid this?" becomes a question with a framework behind it rather than a negotiation, and a manager can answer it without checking with you first. That shift is most of what compensation and rewards strategy work delivers in a small business — not a repricing, but a defensible answer.

Using it at the point of hire

The structure earns its keep at offer time. A role is mapped to a grade before it is advertised, the offer sits within the band, and anything outside needs a named approval. Without that rule, every tight hiring month reintroduces the problem you have just fixed.

It also makes the offer conversation better. A candidate told where the offer sits in a range, and what progression within the grade looks like, hears something more durable than a number — and is less likely to reconsider at home, which is where a surprising share of recruitment process failures actually occur.

When to buy market data

Once the internal structure exists and you have a specific question: are we losing people in one grade because our band is below market, or for another reason? Data bought before the structure exists usually gets applied unevenly and recreates the inconsistency in a more expensive form.

And where the answer turns out to be that pay is not the cause — which it often is — the structure has still done its job, because it has removed pay as an explanation and pointed attention somewhere more useful. That is usually the manager, the progression path or the work itself, which is where compensation and rewards strategy work stops and other conversations begin.

What to do in the first cycle

Do not announce a new structure and apply it in the same month. The sequence that works is quieter: build the grades, map everybody, identify outliers, and use the structure to govern the next increment round without making the round itself an announcement about grading.

By the second cycle the structure is visible in how decisions are being made, and explaining it is straightforward because people have already experienced it working. Announced first and applied later, it invites every employee to work out their own position before you have finished thinking about the outliers.

What to do when someone asks which band they are in

Bands exist for weeks before the question arrives, and how it is answered sets whether the structure survives.

The instinct is to be vague, because the honest answer sometimes places a person lower than they expect. Vagueness is worse than the answer. A structure nobody will discuss is read as a structure with something to hide, and people conclude their position is worse than it is.

The workable response has three parts: the band, what distinguishes it from the one above, and what would have to be true to move. The second part is where most businesses have not done the work, and it is the part employees actually want. Being told you are in band three is information; being told what band four requires is a conversation about the future.

Expect a proportion of people to disagree with their placement, and treat disagreement as useful rather than as a problem. Where several people in similar roles all believe they are placed too low, the band descriptions are probably wrong rather than the employees. Where one person does, it is usually a conversation about performance that was overdue — and the banding exercise has surfaced it rather than caused it.

One practical warning. The moment grades exist, somebody will ask which grade they are in and what it takes to reach the next one. Have an answer before you start, because the second half of that question is a progression question rather than a pay one — and answering it well means knowing what each role actually owns, which is role design and job descriptions territory rather than compensation.