The Smaller the Company, the Rarer the Raise. It's Not a Budget Problem.

By Kris Brown, SPHR — Founder & Principal, Capra HR

At companies with fewer than 50 employees, 56% of workers got no raise in the past year. At 50 to 499 employees, it's 31%. At 500 and up, 27%. That's from BambooHR's 2026 compensation survey of 1,500 US salaried employees, and it's the cleanest picture I've seen of a pattern I run into constantly: the smaller the company, the rarer the raise.

Same labor market. Same inflation. Same employees reading the same salary data online. So why does the raise rate collapse as headcount shrinks?

It's not that small companies are broke, and it's not that owners are cheap. In my experience it's simpler and more fixable than either: big companies have a machine that runs raises, and small companies have a busy owner and no trigger. Nothing on the calendar ever says "decide pay now." So pay gets decided by whoever forces the question.

No trigger, no raise

Here's the mechanism. Without a compensation structure, a raise happens in exactly three situations: an employee works up the nerve to ask, an employee shows up with a competing offer, or the owner has a guilt spasm after a great quarter. Every one of those is a negotiation. Negotiations are uncomfortable. So everyone — owner included — quietly avoids the topic, and the default outcome is nothing.

The market doesn't avoid the topic. In that same BambooHR data, two-thirds of people who changed jobs got a raise on the way in, and those raises averaged 15.3%. People who stayed put averaged 4.8%. Your competitor doesn't need to be a better employer than you. They just need to be the only one who made an offer.

And the ask-or-nothing system has a nasty side effect: it pays the squeaky wheel. The employees who negotiate hard get raises. The quiet, conscientious people who assume good work speaks for itself get nothing, and those are usually the people you can least afford to lose. You're not underpaying everyone. You're underpaying the people who don't ask.

The bill arrives anyway

The money you're not spending on raises shows up anyway, in four forms.

Pay compression. Market rates for new hires keep climbing while incumbent pay sits still, so eventually a new hire out-earns the loyal person training them. In beqom's 2025 survey, 32% of employees said new hires at their company are paid more than experienced, tenured staff. And they find out. Pay comes up at lunch, in job postings, on Glassdoor.

Avoidable turnover. In Payscale's 2025 practices survey, 31% of organizations named unfair pay as the top reason they lose talent, and in the BambooHR data, 45% of employees already believe they're underpaid. Gallup's long-standing estimate puts replacement cost at one-half to two times annual salary. Run that math once: a 5% raise on a $70,000 role is $3,500 a year. Replacing that person starts around $35,000.

Counteroffer roulette. When the only reliable trigger for a raise is a resignation, you end up doing compensation strategy in the worst possible moment — reactive, emotional, and at the exact market premium you refused to pay proactively. By the time someone hands you an offer letter, the relationship damage is usually done.

Transparency laws. Illinois, Minnesota, Massachusetts, New Jersey, and Vermont all switched on salary-range disclosure requirements during 2025, and California's tightened rules have been in effect since January. Vermont's law applies from five employees. If you're posting jobs with pay ranges you invented the night before, your own team is reading them — and doing the comparison you never did.

The fix is a page, not a project

When owners hear "compensation structure," they picture what the enterprise HR articles describe: comp committees, benchmarking subscriptions, a consultant, a spreadsheet with forty tabs. At 25 to 300 employees you need almost none of that.

At one client, a support organization had accumulated 60 job titles with no consistent leveling. Pay decisions were unmanageable, and unfair in ways nobody had chosen on purpose. We consolidated those 60 titles into 15 levels, repriced every role, and built the business case that brought more than 20 underpaid people back within range. The hard part wasn't the design. The hard part was that nobody had ever made it somebody's job.

The minimum viable version fits on one page:

• Levels. Three to five per function. Not 60 titles — a handful of levels that describe real differences in scope and skill.
• A range per level. Informed by the market data you can actually get (free sources are fine to start), honest about being approximate. A rough range beats no range every time.
• A raise calendar. One comp review per year, on a date everyone knows, plus a defined exception path for promotions and retention emergencies. This is the piece every template skips, and it's the whole cure: the review happens because the calendar says so, not because someone threatened to quit.
• One sentence of philosophy. Where do you aim to pay, and what do you reward? Write it down so decisions stop depending on the owner's mood.

Then the part no template covers: the transition. Building structure exposes the drift you already have, and you owe the people in it straight answers. The long-tenured person who's underpaid gets fixed fast, this year, not phased over three, with a plain acknowledgment. The person above range gets an honest freeze, never a cut. Budget the fixes first; if you can't fix everything at once, fix the worst gaps and tell people the sequence. Silence is what breaks trust, not the gap itself.

Where to start

Week 1: One spreadsheet — everyone's role, tenure, and pay, sorted by role. The outliers will be obvious in ten minutes. That discomfort you feel is the project justifying itself.
Weeks 2–3: Draft the one-pager — levels, ranges, the annual review date, the philosophy sentence.
Week 4: Fix the two or three worst gaps, and announce the calendar: here's when comp gets reviewed, here's how decisions get made.

You don't need a raise budget the size of an enterprise's. You need a trigger that isn't a resignation letter.

If you'd rather not build it alone, this is exactly the kind of project I do.