What should I do with my life? A 4-question diagnostic

The search query is malformed. That's why you can't find an answer. "What should I do with my life" wants one noun back (a job title, a calling, a verb you can build a Tuesday around), and real lives don't reduce to one noun. They run on four or five overlapping commitments that re-weight every seven or eight years. So you keep getting nothing back, and it isn't your fault.

There's no answer to "what should I do with my life" in the form you typed it. Replace it with four narrower questions — money-no-constraint, child-self skill, specific envy, conversational return — and look for the overlap. Where three of four converge is your direction. Where they don't is also data.

The next twenty minutes give you the four sub-questions, how to score them, the three ways the exercise breaks, and what to do with what you find.

If what you're really asking is whether to leave a job that pays well, that's a different piece. See career change at 40; that one's the runway math. This one is the question underneath.

Why "what should I do with my life" is the wrong question (and the better one underneath)

Three problems with the original question.

It's a single-output question pretending to be open. Your brain reads it as: give me one job title. The implicit grammar wants a noun. But people don't have a "what to do with life" the way a fork has a use. They have a portfolio (paid work, primary relationships, a creative practice, civic involvement, care of older parents), and the portfolio shifts.

It assumes the answerer has the information. You don't, yet. The four questions below are scaffolds for generating the information from your own life, not for retrieving an answer already inside you.

It runs on the wrong time horizon. "Life" is too long to be operative. The honest planning horizon for most knowledge workers in their late 30s and early 40s is closer to five years. Past that, the world changes faster than the plan.

A better question, in writing: given my current obligations, what are the four narrowest answerable sub-questions that triangulate a direction for the next five years?

The 4-question diagnostic

This is the closest thing to a "what should I do with my life quiz" I trust, and it's not a quiz. Each of the four questions has a strict form. You answer in writing, not in your head. Sit with each one for ten minutes minimum before moving on. One specific answer per question.

Q1. The money-no-constraint question

If money were genuinely no object, $5M in a brokerage account drawing 4% covering your life forever, what would you do with the next five years? Not "travel." Not "be happy." Be specific. First month, first year, third year.

The trap: most people answer Q1 with leisure, then second-guess. Push past it. Almost nobody actually wants thirteen years of leisure. The second and third answers are the ones to write down.

What you're looking for: the activity you'd structure your weeks around. Not the consumption pattern.

Q2. The child-self skill question

What did you do, between roughly ages 8 and 14, that absorbed you for hours at a time without anyone making you? Not the activity that made your parents proud. Not the one you were praised for. The one that pulled you in.

Be granular. "Reading" isn't granular enough. "Reading mystery novels and writing my own bad versions in a spiral notebook" is. "Sports" isn't granular enough. "Organizing the neighborhood kids into a stickball league with a written schedule and standings" is.

Why this matters: a useful frame from adult-development psychology — the kind of work Mihaly Csikszentmihalyi (flow) and Howard Gardner (Multiple Intelligences, originally 1983, revised 2011) anchor — is that the cognitive shape of an adult's engaged work tends to show up early in childhood absorption patterns. Not the topic. The cognitive shape: pattern-spotting, narrative-building, system-organizing, materials-making. Q2 is fishing for that shape, not for the literal hobby.

Q3. The specific-envy question

Whose career, name a real person you know personally, produces a specific twinge of envy when you hear them describe their week? Not "Bezos." A real person, in your life, doing real work.

Now narrow the envy. It's almost never the whole job. It's a slice. The autonomy over Tuesdays. The collaborator she works with. The fact that he leaves at 4:30. The specific kind of problem she solves before lunch. The type of conversation he has at work.

That slice is your signal. Your specific envy is a high-fidelity instrument because it's involuntary. You can't will it. You also can't fake it; if no slice is producing envy, that too is data (see §6 below).

Adam Grant has written extensively on envy as career-information, most accessibly in Think Again (2021) and his Wharton work. The framing predates him, going back to Aristotle's Nicomachean Ethics Book II. The operational use of envy as instrument rather than vice is what we want here.

Q4. The conversational-return question

At a dinner party, or the 2026 hybrid equivalent (a friend's kitchen on Saturday morning), what topic do you keep returning to whether or not the table is interested? What do you bring up unprompted?

Three conditions: (1) you bring it up without being asked; (2) you light up when you do; (3) you've done this at least twice in different settings in the last six months.

The trap: confusing what you talk about with what you complain about. Complaining about your CEO is not Q4 material. Spontaneously explaining the failure mode of a product category you don't even work in: that is.

Q4 catches the involuntary curiosity that paid work has usually trained out of you. The honest math: most tier-2 strategy people identify Q4 in about ten minutes. Most VPs at Series-C startups need longer, because the topic-of-the-job has crowded out the topic-of-the-self.

How to score the answers, the convergence test

Write each of the four answers in one sentence on the same page. Underline the operative noun-or-verb in each. Then look for overlap on three dimensions.

Shape overlap. Are the answers about pattern-spotting (Q1, Q2), explaining (Q3, Q4), making things (Q1, Q2), or organizing other people (Q3, Q4)? Cognitive shape is the strongest signal because it's the thing you don't get to choose.

Material overlap. Same domain showing up: health, money, software, language, real estate, food, education, the built environment? Weaker than shape overlap but easier to spot.

Audience overlap. Are you working with kids, with operators, with older adults, with creators, with engineers?

The decision rule: three of four answers must converge on at least one of these three dimensions. Three of four overlapping on shape: strong direction signal. Three of four overlapping on material: useful but weaker; check whether the shape works. All four overlapping is rare; when it happens, check whether you're being honest (see failure modes below).

The convergence answer is rarely a job title. It's usually a sentence like: "the cognitive shape is pattern-spotting; the material is built-environment / cities; the audience is operators rather than consumers." From there, the candidate titles fall out, usually three to seven of them. Then you go test them.

Three failure modes of this exercise

I've watched clients run this badly. Three patterns.

The sunk-cost answer. Q1 reads: "I'd keep doing exactly what I do now, just with less stress." That's almost never the honest Q1. It's the sunk-cost speaking. You've invested eight or twelve years; the brain protects the investment. The corrective: re-answer Q1 with the constraint that your current track is closed, somehow no longer available. What's the second-best life?

The status answer. Q3 reads with envy directed at people whose career has high external status: the friend who made partner, the cousin running a fund. That's envy of the credential, not envy of the slice. Real Q3 envy points at the texture of someone's Tuesday, not their LinkedIn headline. If your Q3 names a job title rather than a slice, you haven't answered yet.

The "should" answer. Q4 reads: "I keep bringing up climate change." You should care about climate. Most thoughtful people do. But if it's the should-care speaking rather than the unprompted-curiosity, Q4 is contaminated. The test: would you talk about it at length to someone who already shares your position? Real Q4 topics survive the in-group test. "Should" topics die there.

What to do with the answer

Three branches, depending on what convergence looks like.

High convergence (3-4 of 4 align on shape or material). Write the sentence. Generate three to seven candidate job titles. Identify one person, in your network or two degrees out, doing each one. Have a 30-minute conversation with at least three. The script: "I'm running a structured exercise on direction. Your week looks like one of my answers. Can I ask what surprised you about it?" Most people say yes.

After three conversations you'll have either eliminated all the candidates (shape signal wrong; back to the exercise), narrowed to one or two, or most often discovered a fourth title you hadn't generated. Then it's the runway-and-pivot work, which I wrote up in career change at 40. That's the math piece.

Partial convergence (2 of 4 align). Common, and actually fine. Two answers in the same direction is enough to run a 90-day experiment without quitting anything. Pick a small version of the candidate direction (one client project, one weekend course, one volunteer slot) and do it for three months. The two non-converging answers usually clarify in light of the experiment.

Zero or one of four align. The least common, and the most useful result, even though it doesn't feel that way. Zero convergence usually means you don't have enough material yet. You've spent so long in your current track that your child-self skill is hard to remember, your specific-envy is muted, your conversational return is your job, and your money-no-constraint answer is "rest." Not solvable in a search bar. Give yourself six months of cheap input. Read in three domains you've never read. Meet five people in jobs you don't understand. Run a small failure on purpose. Then re-run the four questions.

If zero convergence persists after six months of new input, the question underneath is probably not vocational. See below.

When the honest answer is "I don't know," and why that's also data

About a quarter of clients come back at week two and say some version of: I ran it. I sat with each question for the full ten minutes. I'm still not sure. Most of the time this is one of three things, and the response is different for each.

It can be early-season fog. You're three months out of a hard transition: a death, a divorce, a move, a startup folding. Vocational signal is genuinely down because your nervous system is allocating cycles to the larger event. Don't grind harder on the four questions. Give the system another quarter. Real signal usually comes back around the six-month mark.

It can be chronic-overconsumption fog. You've read seventeen versions of this article in the last year. You're well-informed about other people's frameworks and starving of your own data. Unfashionable corrective: stop reading career-direction content for ninety days. Do one new thing a week (a class, a side project, a conversation in an unfamiliar room). Direction-signal needs new inputs, not new frameworks.

And it can be the deeper question. "What is my purpose" or "what should my life mean" is a different question from "what should I do with my life," with different tools. Viktor Frankl's Man's Search for Meaning (1946 original; 2006 Beacon Press English edition) addresses it directly. So does some of Esther Perel's work on integration of identity and work. When the four questions don't move and there's no obvious external reason for the stall, the question underneath is often meaning, not vocation. Different problem. Different tool.

So: pen, paper, the four questions, ten minutes each. No phone. Don't write a good answer; write the true one. The exercise takes about an hour properly done. Most people skip it and read more articles instead. The cost of trying is one notebook and a Sunday morning. Most clients land somewhere usable inside the hour. If you don't, you got data; try again next week.

— Maren