Software Engineers: You're Probably Switching Jobs for the Wrong Reason

In February 2026, ADP's pay tracking data showed the pay premium for switching employers hit a record low since tracking began in 2020. Job stayers saw wages rise 4.5%. Job switchers saw 6.3%. A gap of less than two percentage points. At the peak of the Great Resignation in April 2022, that same gap was 16 points.

The conventional wisdom for software engineers: stay two years, then jump for 20-30%. It worked. Then the market changed, the advice didn't, and a generation of engineers kept job-hopping like it was still 2021 while the premium shrank to basically a rounding error.

The deeper problem: salary was always the wrong metric to trigger the decision. The engineers who time their moves well aren't watching their pay stubs. They're watching their learning curves.

Salary Is a Confirmation Signal, Not a Trigger

Your internal comp falls behind the market before you notice it. Companies cap merit increases at 3-10% per year. The market moves in bigger steps. By the time the gap is obvious enough to make you angry, you've probably been underpaid for a year and a half.

The pay gap should confirm a decision you've already made, not initiate one.

Use Levels.fyi and peer conversations to check your market rate annually. If you're materially behind, that's confirmation. But the decision to start looking should come from something earlier: whether you're still accumulating career capital at your current company. If both answers are fine, you're not in a hurry. If neither answer is fine, you're already late.

External Hires Get Paid More and Perform Worse

Wharton professor Matthew Bidwell studied thousands of promotions and external hires at a large financial services firm. Companies pay external hires 18-20% more than internally promoted employees for the same roles. You knew that part. It's the reason your recruiter DMs feel like a daily affirmation.

The twist: external hires receive significantly lower performance evaluations for their first two years. They also exit at a much higher rate during that adjustment period, both fired and voluntarily.

You aren't comparing Current You to New Job You on day one. You're comparing Current You to the version of yourself 24 months into an unfamiliar organization.

Bidwell's conclusion is blunt: "Skills are much less portable than you think they are." The social capital you've built, the context you carry, the relationships that make things actually move. None of that transfers. You're spending the salary bump to buy back ground you already had. It's not a raise. It's a round-trip ticket to zero.

None of this means never switch. It means when you switch, you need a reason worth a two-year penalty. "It pays more" doesn't clear the bar. "The scope will push me into skills I can't get here" does.

Your Learning Curve Is the Real Clock

Cal Newport's career capital framework and Charles Handy's Second Curve theory both arrive at the same place: the optimal time to move is while you're still ascending the first curve, not after you've spent two years at the plateau.

The plateau is easy to recognize in retrospect and hard to recognize in the moment. Alex Hyett, a 12-year engineering veteran, wrote about spending two years in a role where he couldn't count the new technologies he'd learned on one hand. The work was fine. He was competent. He was stagnant. He calls those two years his biggest career regret. The salary was comfortable. The comfort was the trap.

Ask yourself once a year: can you name five specific skills or areas where you'll be meaningfully better 12 months from now if you stay? Not "I'll get more experience." Specific capabilities. If you're hedging, you have your answer.

Most developers hit this wall around years four to six. The danger isn't noticing the plateau. It's noticing it, feeling vaguely uneasy for 18 months, and doing nothing because switching is hard and the salary is fine and there's always next quarter.

The Equity Math Most Engineers Skip

RSU mechanics reward patience, but only up to a point. And most engineers do not actually run the math.

The classic 4-year vesting schedule (25%/25%/25%/25%) makes year two look identical to year four. But big tech increasingly uses front-loaded schedules: 40/30/20/10. At Google and Nvidia, you vest more in years one and two than in years three and four. After year three, refresh grants drive your total comp, and those are performance-dependent and discretionary.

Two distinct mistakes follow from ignoring this.

Leaving eight weeks before a one-year cliff because you got an exciting offer, forfeiting 25% of your initial grant. A sign-on bonus rarely covers it. Do the math before you accept.

Staying through year four of a flat schedule, fully vested, then letting inertia keep you another eighteen months because the equity felt safe. It doesn't feel safe. It's already gone.

Compare three-year total comp, not first-year packages. Stack your unvested equity at the current company, the new offer's vesting schedule, and the sign-on. Then subtract the fact that you'll underperform for two years. The math is often closer than it looks. And once you've run it and decided to move, how you counter the offer determines whether you capture the full market premium.

The Option Nobody Actually Takes

Internal lateral moves are dramatically underutilized. The internal mobility rate across most tech companies sits at around 8%. That 8% gets promoted faster and sees higher long-term pay growth than people who either stay in the same role or switch companies.

The logic is almost offensively simple. You keep your social capital, your organizational context, your unvested equity, and your performance track record. You get a new learning curve. You avoid the two-year adaptation penalty. Everyone wins.

Before you open a new tab for job listings, spend three months genuinely trying to engineer an internal move. Talk to your manager. Talk to teams you're curious about. If the company can't give you a new challenge within six months, that tells you something real about the ceiling. Now you have signal, not just restlessness.

A lot of engineers leave to get the thing they could have gotten by asking. They just didn't ask because asking felt awkward. Awkward five-minute conversation now, or a two-year penalty on the other side. Seems like a reasonable tradeoff.

When Should a Software Engineer Switch Jobs? All Three Tests Have to Pass

Not signals to watch. Tests to pass. All three.

First: your learning curve has peaked, and you've already tried the internal fix. Not "I'm bored with my tasks." The bar is whether you can project meaningful skill growth forward and whether you've actually had the conversation with your manager about changing your scope. If both answers are no, you haven't done your homework yet.

Second: your comp is 20% or more below your market rate, and the path to close it internally is blocked. If you've surfaced the gap with your manager and they've confirmed the band ceiling won't move in the next two review cycles, that's a real constraint. A 20% gap compounds. But 8% or 10% is often recoverable internally faster than you think, especially if you're performing at level. Understanding how software engineer levels and comp bands work helps you know whether that gap is structural or a negotiation problem.

Third: your manager is the ceiling, not the company. A 2026 Sagepub study found that manager quality explains 22% of the variance in career development outcomes. A manager who misrepresents your work upward, blocks your visibility with leadership, or consistently underlevels your contributions is a structural constraint on your trajectory. A lateral move sometimes solves it. Leaving does too.

Two out of three isn't enough. The cases where switching clearly wins are where multiple constraints stack simultaneously. One bad thing is a signal. Three bad things is a verdict.

Don't Wait for Misery to Motivate You

You shouldn't need to feel miserable to know whether the market sees you as worth more than your current company pays. Interview externally twice a year, not to get offers, but to stay calibrated. Run one or two full loops per year. If you're reaching final rounds consistently and the numbers are surprising, you have information. If you're being filtered out early, you have different information. Both are useful.

This is also how you show up prepared when a real opportunity appears, instead of scrambling to relearn interview mechanics under time pressure. At SpaceComplexity, the voice-based interview practice is built for exactly this: staying sharp so you can move when it's right, not panic-practice when you already have an offer deadline.

If you're in year three or four, run your equity math quarterly. The answer changes as the stock price and refresh schedule change. Every twelve months, answer the learning question honestly. Five specific skills. If you're hedging, you have your answer.

Three Scenarios, Three Calls

Year two. Learning. Internal comp 10% behind market. Stay, raise the gap with your manager, give it one review cycle.
Year four. Learning curve peaked 18 months ago. Comp 25% behind. Talked to two internal teams, neither had headcount. A real case for moving.
Year three. Learning a lot. Got an offer at 30% more base. $120K unvested over the next 14 months. Run the equity math. The offer needs to cover the forfeiture plus the two-year adaptation discount. Maybe it does. Maybe it doesn't. Don't guess.