You should leave when at least three of these are true for at least three months: compensation 15%+ below market, no new skill or scope in the last 90 days, a manager you don't trust, dread on Sunday night, no clear promotion path, or work that's actively damaging your health. One on its own is fixable. Three at once is a pattern, and patterns are decisions waiting to be made.
The median engineering tenure in 2026 is about 2 years 11 months — not because everyone is restless, but because the moment growth stops, the cost of staying compounds faster than the cost of leaving.
Most "should I quit?" articles are written from one of two extremes. One half tells you to grind it out, keep your head down, be loyal. The other half tells you life is short and your dream job is one resignation letter away. Neither is honest about the part in the middle — where you're not miserable, but you're not growing, and you've been telling yourself "I'll give it six more months" for two years.
This is the field guide for that middle. Twelve signals, each with the test that tells you whether it's noise or a pattern. Then the decision framework — including the one thing you should always do before you accept an offer or hit send on your resignation.
The 12 signals
Loyalty is admirable. Loyalty discounts on your paycheck are not. If a recruiter could place you in a comparable role tomorrow for 15–25% more, you're paying a tax to stay. That tax compounds: it shows up at promotion time (your raise is calculated from a low base), at job-change time (the next offer is anchored to your current comp), and at retirement (every dollar today is a dollar plus market returns in 20 years).
If your next 90 days are functionally identical to your last 90 days, you're done learning here. Some plateaus are fine — mastery is a real thing, and there's a season for going deep before you go wide. But if you've been in the plateau for more than two quarters and there's no plan to break out, the company has cast you in a role and you've accepted the part.
People don't leave companies, they leave managers — this cliché is exhausting because it keeps being true. A bad manager can poison a great team, a great mission, and great comp. The hard part is that "bad" doesn't always mean cruel. It can mean absent. Unclear. Conflict-averse. Politically captured. Unable to fight for you in promotion calibrations.
Everyone has a Monday they wish they could skip. But if Sunday afternoons have a specific weight to them — the texts get shorter, the wine gets larger, you find yourself "just checking Slack" at 9pm to get ahead — your nervous system is telling you something your professional brain is still arguing with.
This one is sneaky because the cause is rarely the work itself. It's usually one person, one project, or one specific looming conversation. Identify what you're actually dreading. Sometimes that conversation, once had, fixes the whole problem.
This one usually arrives before the others and goes unnoticed. The big project you used to lead is now co-led with someone newer. The architecture decision you would have made now gets routed up. You're being managed out one responsibility at a time, often by a manager who doesn't yet know they're doing it.
You've had the "what does it take to get to the next level?" conversation. The answer was vague the first time and vague the second time. The colleague who joined six months ago and got promoted last cycle had the same scope you have now. The leveling rubric exists on Notion but no one in your management chain can articulate which line item you're failing.
The strategy changed. The product pivoted. The thing that pulled you in — the customer, the technology, the market — is no longer the company's center of gravity. You can stay and adapt, and many people do. But if you've spent the last two all-hands feeling like an alumnus, that's not loyalty fading. That's the company quietly becoming a different company than the one that hired you.
This sounds like flattery and feels like a trap. If everyone you used to learn from has left, and the people replacing them are at your level or below, your ceiling has dropped. The people who would have stretched you into the next level are no longer here to do it. You can train others, which is valuable. But you stop being trained, which is the part that compounds.
Every company has an aspirational set of values painted in a meeting room. The interesting question is what gets rewarded when those values conflict with shipping faster, hitting a number, or protecting an executive. If "we value transparency" produces an all-hands where no one will say what they actually think, the value is dead and you're being asked to pretend it isn't. That pretending is exhausting in a way that doesn't show up on any timesheet.
Sleep is shorter. The Sunday headache. The recurring stomach thing your doctor has no explanation for. The drink you didn't used to need. The relationship that's started absorbing the spillover. If any of these arrived in the last 6–12 months and they map to a change at work, your body has already filed its complaint. The job will adapt to whatever you can take. Your nervous system won't.
Five years ago, half your peers were at companies you respected. Today, the same peers have all moved to better places, and you're the one still defending the org chart. This isn't FOMO — it's signal. If the smartest people you know have rotated out and you're still here, ask why. Often the answer is comfort, and comfort is the most expensive thing you can buy with your career.
Here's the simplest one. Imagine the company calls you tomorrow and says: "We're not hiring externally for your team this year. If you want to keep your job, you have to re-interview against the same bar we'd use for an external candidate. Same role, same comp." Would you choose to re-interview? Would you put in the prep? Would you take the offer?
If the honest answer is no — you'd let the role go and look elsewhere — you don't actually want this job. You just don't want the friction of leaving it. Inertia is the most expensive feature of your career.
The "six more months" test
Most people don't quit too early. They quit two years too late. The reason is almost always the same: the cost of staying is invisible until you leave, and the cost of leaving is visible every day until you do.
The honest test is this: forget what you'd be giving up. Ask what you'd be gaining. If you stay six more months, what is materially different on day 181? A new project? A title? A real raise? A specific skill? If the answer is "things will probably stabilize" or "the new VP might fix things," that's a wish, not a plan. Wishes don't earn equity.
Before you decide, do this
Whether you're leaning stay or leaning leave, run this three-step exercise. Most people skip it. The ones who do it almost always make a better decision than the ones who don't.
- Write the offer letter you'd want. Title, scope, comp, team, manager, learning. Don't filter. This is the bar your current job needs to meet or get out of the way of.
- Take one real interview. Not three. One. Not to leave — to recalibrate. Most people discover they're 20% underpaid or 20% under-titled within a single loop. That data changes the conversation with your current manager. Browse roles on the live job board to start.
- Have the honest 30-minute conversation with your manager. Not a vague "I'm thinking about my career." A specific: "Here's what I want in the next 12 months. Is that available here?" If yes, you have a path. If no, you have your answer.
If you decide to leave
The tactical stuff matters more than people admit. Give two to three weeks notice — not one, not four. Write the transition docs no one asked you to write; they are the cheapest reputation capital you'll ever buy. Tell your manager before you tell the team, and tell the team before they hear it through the grapevine.
Do not badmouth on the way out. The tech world is a small town pretending to be a city. The mid-level engineer you vent about today is the VP of engineering screening your resume in 2031. The recruiter you ghost is at four companies in the next decade. Treat the last two weeks as the first two weeks of the next chapter, because that's what they are.
And before you accept the next offer, do the thing most people don't: research the culture. A 25% raise into a place with a 2.8 work-life-balance score is not a raise. Browse the culture directory to read what actual employees say about the teams you're considering. Compensation can be negotiated. Culture is the part you're stuck with.
If you decide to stay
Staying isn't passive. If you choose to stay after honestly running these signals, the choice has to come with a plan. Write down what would have to change in the next 90 days for the decision to be the right one. If those changes don't happen, the next 90 days aren't a re-evaluation — they're a notice period you forgot to file.
The worst outcome isn't leaving a job too early. It isn't even staying too long. It's the quiet drift, where you wake up three years from now in a job you would never accept today, with skills you stopped sharpening in 2026, at a comp that lost a bidding war you never knew you were in.
Trust the pattern. Run the test. Then act.
Browse roles at companies people actually like
Every company on JobsByCulture is profiled with real Glassdoor data, work-life-balance scores, and culture values evidenced by employee reviews. No marketing copy — just signal.
Browse Open Roles → Take the culture quiz →