How to Set Employee Pay (And Handle Raise Requests Without Guessing)

Almost every small business sets pay the same way. Someone good is about to walk, or a candidate names a number, and you decide in the moment what you can live with. It works. You keep the person. You move on.

Do that eight or ten times over a few years and you wake up with a payroll nobody can explain. Two people doing the same job are paid meaningfully differently because one of them was hired during a panic. The employee who never asks for anything is the furthest behind. The one who asks every spring is the closest to the top. And you can't fix any of it, because every number is a promise you already made.

The problem was never generosity or stinginess. It's that pay got decided one negotiation at a time, and negotiation rewards the willingness to negotiate — not the value of the work.

Decide the Structure Before You Need It

The fix is unglamorous and takes an afternoon: decide what each role is worth before you're sitting across from a person asking.

Start by listing every role in the business, not every person. Roles, because you're pricing jobs, not personalities. For each one, build a band with three points: a bottom for someone who can do most of the job with support, a midpoint for someone fully capable and independent, and a top for someone who does the job at a level you'd struggle to replace.

To set those numbers, triangulate. Public salary surveys give you a national baseline. Your industry association almost certainly publishes compensation data for your sector — that's usually the most relevant source and the most underused. Then look at live job postings for comparable roles in your actual region, because that's the market a candidate is really choosing between. Where those three sources cluster is your midpoint.

Two rules keep the bands honest. Bands should overlap between adjacent levels, so a senior person early in a level and a strong person topping out a level below can be paid similarly without breaking the logic. And the top of a band should be a number you could actually pay a great performer without going back to the drawing board — if it isn't, it's decoration.

Then Check It Against What the Business Can Carry

Market data tells you what a role costs. It doesn't tell you what you can afford, and confusing the two is how a business ends up with a payroll it can only support in good months.

Run the whole thing as a percentage of revenue. Total payroll, fully loaded — wages plus payroll taxes, benefits, workers' comp, and your own compensation — divided by revenue. There's no universal correct figure; a services firm and a distributor live in completely different ranges. What matters is that you know your number, that you know what's normal for your industry, and that you know what happens to it in a slow quarter.

Then do the part most owners skip: model the bands at full cost. Not what you pay today — what you'd pay if everyone were sitting at their band midpoint. If that number breaks the business, your bands are aspirational, and you'll discover it the hard way in about eighteen months.

A pay structure you can only afford in your best year isn't a pay structure. It's a promise you'll have to break.

Place People Honestly, Including the Awkward Ones

Now put every current employee into a band. This is the uncomfortable hour, and it's the one that pays for itself.

You'll find three situations. Some people are placed about right, and you can stop thinking about them. Some are below their band — usually loyal, usually quiet, usually the ones who never asked. And occasionally someone sits above the top of their band, because a hiring panic set their number.

The people who are underpaid relative to the work are your urgent problem, not your generous gesture. They are the most likely to leave, the most expensive to replace, and the ones whose departure will make you say "I had no idea they were unhappy." You had the information. It was on this page.

You probably can't correct everyone at once. So sequence it: biggest gap first, then flight risk, then tenure. Tell the people you're correcting that you reviewed pay across the business and are adjusting theirs to match the market for their role. Don't dress it up as a reward for performance — it isn't one, and calling it that means you'll have to invent another reward later.

For someone above their band, don't cut pay. Freeze the base, let the band grow past them over time, and route future increases through bonuses tied to results. It resolves in two or three years without a betrayal.

Put Raises on a Calendar Instead of a Conversation

Once bands exist, raises stop being a negotiation and become a schedule. Pick one review cycle a year, put it on the calendar, and hold it whether or not anyone asks. Reviewing everyone at the same time forces you to compare people against each other and against the band — which is exactly the comparison that ad-hoc raises let you avoid.

Separate the two kinds of increase, because they answer different questions:

• Movement within a band reflects growing capability in the same job. It's incremental and it's earned by performance.
• Movement to a new band reflects a genuinely bigger job — more scope, more responsibility, more judgment. It's a step, not a bump, and the job has to actually change first.

Say this out loud to your team, because it prevents the single most common misunderstanding in small businesses: someone doing their existing job very well for four years assumes that alone should move them toward the top of the market. Within-band, it should. Into a new band, it shouldn't — and hearing that in year one is far better than hearing it in year four.

Also decide in advance what happens when there's nothing to give. A year comes when the business can't fund increases. That's survivable if you say it early, explain the number behind it, and tell people what would have to be true to change it. It is not survivable if people find out by watching the review date pass in silence.

The Raise Conversation, Handled

Someone will still walk into your office off-cycle and ask. Here's the version that goes well.

Don't answer in the room. Not yes, not no. Both are mistakes made under pressure. Thank them for raising it directly, ask what number they have in mind and what it's based on, and take notes.

Set a date and keep it. "Give me until Friday" is a complete answer. Vanishing for three weeks is what turns a pay question into a job search.

Answer against the structure, not the mood. Where are they in their band? Where would the raise put them? Is the job actually bigger, or the same job done well? The structure gives you a defensible answer instead of a feeling.

If it's yes, be specific. The new number, the effective date, and what it reflects. Don't undersell it by mumbling.

If it's no, make it a real no with a path. "Not right now" plus three concrete things that would change the answer plus a date to revisit. People can work with an honest no. What they can't work with is a maybe that never resolves — that's the thing that quietly turns a good employee into a candidate.

Say More About Structure Than You Think You Should

Most owners are more secretive about pay than the situation warrants. You don't have to publish individual salaries. But you should tell your team how pay is determined: that roles have bands, that bands come from market data, that raises are reviewed on a set cycle, and what distinguishes a within-band increase from a move to a new band.

The reason is simple. Your employees discuss pay with each other regardless of your preferences, and in a business with a handful of people they will eventually compare notes. When they do, a difference either lands as a level or as favoritism — and which one it lands as depends entirely on whether there's a visible structure behind it.

Where Owners Get This Wrong

A few failure modes account for most of the damage.

Paying for loyalty instead of the role. Tenure is real, and it belongs inside the band, not above it. Someone who has been with you a long time doing an entry-level job is still doing an entry-level job.

Counter-offering to stop a resignation. It usually buys months, not years, and it teaches everyone watching that the fastest route to market pay is another offer. If someone is genuinely underpaid, fix it before they're holding an envelope.

Confusing what you can pay with what you should. A good year is not a permanent raise in disguise. Increases stay on payroll after the good year ends; that's what bonuses are for.

Deciding alone. This is the quiet one. Compensation is the decision where an owner's blind spots are widest, because you're pricing people you like, in a market you don't shop in, using a budget you feel personally responsible for. Nearly every owner underestimates what a role now costs and overestimates how satisfied a long-tenured employee is with a number set three years ago.

That's exactly the kind of judgment that improves enormously with one outside read. Not someone who knows your team — someone with no stake in the outcome who will look at your bands, your payroll percentage, and your placements and ask what you'd have to believe for all of it to be right. Then hold you to the review date you set, which is the part that actually gets skipped.

A Two-Hour Version

If you do nothing else, do this in one sitting:

1. List the roles in your business, not the people.
2. Build a band for each — bottom, midpoint, top — from industry survey data and live local postings.
3. Run the math: fully loaded payroll as a percentage of revenue today, and at full band cost.
4. Place every person and mark who's below their band.
5. Sequence the corrections by size of gap and risk of loss, with dates attached.
6. Pick the review month and put it on next year's calendar before you close the file.

None of that is hard. It's just work that never becomes urgent until the day it becomes expensive — which arrives as a resignation letter from the person you assumed was fine.

Frequently Asked Questions

How do I decide what to pay a new employee at a small business?

Set the range before you meet anyone. Triangulate market data for the role in your region — public salary surveys, your industry association's compensation report, and live postings for comparable jobs — then define a bottom, midpoint, and top. Hire near the bottom when the person needs development, near the midpoint when they can do the job today, near the top only when they bring something the role genuinely requires. Fixing the range in advance is what stops a strong negotiator from resetting your whole structure in one conversation.

What should I say when an employee asks for a raise I can't afford right now?

Don't answer in the room, either way. Thank them, ask what number they have in mind and what it's based on, and commit to a specific date for your answer — then keep it. If the answer is no, say plainly that the business can't support it right now, name the concrete conditions that would change that, and set a review date. An honest no with a path attached keeps people. A vague maybe is what starts the job search.

Should small businesses be transparent about pay ranges?

You rarely need to publish individual salaries, but you should share the structure — how pay is set, what bands exist, and when raises are reviewed. Your team compares numbers whether you approve or not. With a visible structure behind them, differences read as levels. Without one, the same differences read as favoritism, and that impression is much harder to undo than it was to prevent.