Salaries stop growing because companies price new hires against the market but price existing employees against last year’s paycheck, and those two numbers drift apart by 10–20% within three years. The plateau is not a verdict on your performance: it is a structural bet that you won’t check the market, and it only breaks when you change one of the four anchors your pay hangs on (the role, the level, the company, or the skill you’re priced for).
That’s the whole answer in two sentences. The rest of this post is the mechanism, the math, and the fix.
The mechanics: salary bands and pay compression
Every mid-sized or large employer runs two pay systems in parallel, and they are not designed to agree.
System one prices candidates. When a company needs to fill a seat, it has to beat the market today. If market rates for your role rose 18% since you joined, the recruiter’s budget rose 18% too, because the alternative is an empty seat.
System two prices employees. Your raise is calculated as a percentage of your current salary, run through a merit matrix, capped by a budget set before anyone looked at your performance. In most companies that budget is 3–4% of payroll. Your “exceeds expectations” rating might get you 4.5% instead of 3%. It will never get you 18%.
Run those two systems side by side for a few years and you get pay compression: the new hire next to you, doing your job with two years less experience, earning more than you. HR knows. Finance knows. It persists because fixing it (repricing every loyal employee to market annually) costs millions, while not fixing it costs only the people who notice. This is one of several hidden rules of pay that nobody explains in onboarding, and it is the single biggest one.
Here is what the gap looks like in practice. Two people start the same role at $70,000. One stays and collects 3% merit raises. One changes employers in years 2 and 4, gaining a typical 15% per move, with 3% raises in between.
| Year | Stayer (3% raises) | Switcher (two moves) | Gap |
|---|---|---|---|
| 1 | $72,100 | $72,100 | $0 |
| 2 | $74,300 | $82,900 (move #1, +15%) | $8,600 |
| 3 | $76,500 | $85,400 | $8,900 |
| 4 | $78,800 | $98,200 (move #2, +15%) | $19,400 |
| 5 | $81,100 | $101,100 | $20,000 |
Five years, identical skills, identical output: the switcher earns 25% more, and every future raise for both of them compounds from those diverged baselines. The stayer never catches up on merit raises alone. Ever.
This is not an argument that you must job-hop. It is an argument that you must know this table exists, because your employer does.
Why working harder doesn’t move a capped band
Here is the part that wastes the most careers: effort and pay are only connected inside a narrow range.
Your salary lives in a band, typically 20–25% wide, attached to your role and level. Performance moves you within the band. Once you reach the top of it, the connection between effort and money is severed. You can work weekends, carry the team, save the launch, and the compensation system will return the same answer: “you’re already at the top of your range.”
The band itself is anchored to four things, and effort is not one of them:
- The role: what job family you’re in, and how close it sits to revenue
- The level: junior, senior, staff, manager, whatever the ladder calls it
- The company: its industry, margins, and compensation philosophy
- The market you’re priced in: which skill set and which talent pool sets your benchmark
Grinding inside a capped band is rowing harder in a docked boat. The oars work fine. The boat is tied to the pier. If your last two raises were 3% despite strong reviews, stop asking “how do I perform better” and start asking “which anchor can I move.”
The four levers that restart growth
Each lever moves one anchor. They are listed cheapest-first: try them roughly in this order.
1. Negotiate (move your position against the band)
The cheapest fix, and the most skipped. Most people have never once asked for a raise with market data in hand; they wait for the review cycle to notice them, and review cycles are built not to. A specific ask (“market rate for this role is $X, here are three sources, I’m asking for a correction to $Y”) succeeds far more often than people expect, because retaining you at market is still cheaper than replacing you at market plus recruiter fees. The scripts, timing, and counters are in our full guide on how to negotiate salary. Negotiation works best when you’re below the band midpoint; if you’re already at the top, it buys one correction, not a new trajectory.
2. Move internally (change the role or level)
If the band is capped, the next dollar lives in the next band. Promotions and internal transfers re-anchor you: a lateral move from a cost center to a revenue-adjacent team often comes with a wider band even at the same level. Internal moves are dramatically easier to win than external ones because you’re a known quantity. Ask your manager one exact question: “What would need to be true for me to reach the next level within a year?” Get the answer in writing. If it’s vague twice in a row, the promotion isn’t coming, and you’ve learned that cheaply.
3. Move externally (change the company)
The nuclear lever, and the most reliable. An external move is the only event where you get priced by system one (market rate for candidates) instead of system two (last year plus 3%). That’s why switchers gain 10–20% per move while stayers collect inflation. It also carries the most risk: new manager, new culture, reset reputation. Two things de-risk it. First, treat the offer stage seriously rather than gratefully. Second, plan the landing: how you spend the first 90 days in a new job determines whether the raise you negotiated becomes a trajectory or a one-off.
4. Reprice your skill (change the market you’re benchmarked in)
The slowest lever and the largest. Sometimes the problem isn’t your employer, it’s the benchmark: the skill you’re priced on has a low ceiling everywhere. Moving from a commoditized skill set to a scarce one changes every future band you’ll ever sit in. Look at how steep the curve is for a software engineer compared with most fields: that steepness is a property of the skill market, not the person. Geography-of-employer is part of this lever too. A developer in Manchester or Lisbon working remotely for a US company earns against a US benchmark while paying local rent; we’ve covered the math in our piece on remote work and global pay arbitrage. Same person, same output, different market, sometimes double the pay.
The 30-minute yearly salary audit
Once a year, calendar reminder and all, run this. Thirty minutes, six steps.
- Pull your real number (5 min). Base plus bonus plus equity, from your last payslip and grant docs. Most people are 10% wrong about their own total comp.
- Find the market rate (10 min). Check job postings with published ranges, our by-country salary data for your role and region, and one or two peers you trust. Three data points minimum.
- Run the comparison (5 min). Put your number through the salary calculator to normalize for location and currency, then use the am I underpaid check to see where you actually sit against the market. Under 5% gap: fine. Over 10%: you have a project.
- Check your band position (5 min). Ask HR or your manager for your band and where you sit in it. In many places (all EU employers under the pay transparency directive, plus a growing list of US states) they must tell you.
- Pick your live lever (3 min). Negotiation, internal move, external move, or skill repricing: which one is realistically available this year? Write it down.
- Book the action (2 min). One concrete step in the calendar: a comp conversation scheduled, one application sent, one certification started. An audit without an action is a mood.
If the honest answer at step 5 is “none, again,” for the second year running, that is the loudest data point the audit can produce.
When staying put is actually right
Job-switching advice gets oversold, so here is the honest other side. Staying is the correct move when:
- You’re below the band midpoint with a strong review coming. Negotiation will likely work; changing companies to fix a fixable gap is overkill.
- A promotion is genuinely in flight. Written criteria, named timeline, sponsor in the room. A level jump plus retention adjustment can match an external move without the risk.
- You’re accumulating something portable. A rare skill, a scarce credential, equity that vests next year. Leaving mid-accumulation sells the asset cheap.
- The intangibles are real money. A four-day week, genuine remote flexibility, a manager who develops you: repurchasing those elsewhere often costs more than a 15% raise pays.
- You’d be catching a falling knife. A 20% raise to join a company doing layoffs in a shrinking sector is not a raise, it’s a severance advance.
The test is simple: staying is right when it’s a decision, not a default. A flat salary is information. It’s your employer telling you, in the only language compensation speaks, how they’ve priced the odds of you leaving. You don’t have to leave. You just have to make that price accurate.