Direction · 8 min read

What to do when you don't know what career you want

Not knowing isn't a personal failing or a phase to get through. It's a normal information problem — and information problems are solved by testing, not by thinking harder.

18 August 2026

Someone asks what you want to do. You say you're not sure. There's a small pause, and in that pause you can feel them deciding you haven't got it together yet. That pause is doing enormous damage to a lot of people, because it frames a completely ordinary situation as a personal deficiency.

You're not behind. You're missing information — specifically, information about what particular kinds of work feel like from the inside. Nobody is born with that. It's collected.

Why thinking about it doesn't work

When you sit and think about your future, you're asking your brain to simulate an experience it has never had, and then predict how you'd feel inside it. We're poor at this. We imagine the job title, the salary and the way it sounds when we say it out loud. We do not imagine the Tuesday afternoon: the actual texture of the hours, which is what determines whether you last.

This is why people can spend two years deciding, commit, and discover in six weeks what a single day of contact would have told them for free.

Start with activities, not job titles

Job titles are a terrible unit to think in. They're bundles — 'doctor' contains diagnosis, paperwork, night shifts, delivering bad news and thirty other things you might love or hate independently. Activities are the honest unit, because they're the thing you'll actually spend your hours doing.

Take ten minutes and write down when you last lost track of time. Not what you're good at, not what looks good — what absorbed you. Then push each one down to the verb underneath it:

• Editing videos for hours → arranging things until the timing feels right.
• Arguing about football tactics → finding the pattern that explains why something works.
• Sorting out a friend's crisis → hearing the real problem under the stated one.
• Rebuilding a bike → making a broken physical system work again.

Those verbs travel. 'Finding the pattern that explains why something works' is sports analytics, epidemiology, fraud detection, product research and a dozen fields nobody has named yet. You can't search for a job title you've never heard of. You can search for the verb.

Widen before you narrow

Most people choose between the five careers they've personally seen — usually the jobs their family does, their teachers do, and television shows. That's a tiny sample of a huge space, and the good fit is statistically unlikely to be inside it.

So before deciding anything, deliberately widen. For each verb, find three directions you'd never considered. Ask someone who works in an adjacent field what jobs exist near theirs that nobody applies for. Read a job board in an industry you know nothing about and note the roles you can't decode. Weird and unfamiliar is the point — the goal at this stage is coverage, not shortlisting.

You can't choose from options you've never seen. Most 'I don't know what I want' is actually 'I haven't met enough of what exists'.

Then test three, cheaply

Pick three directions and run one small experiment each. Small means: finishable this month, costs nothing, needs no permission, and produces something you could describe to another person.

1. 01Have one conversation with someone doing the work. Ask what last Tuesday actually looked like, hour by hour, and what part they'd remove if they could.
2. 02Do a miniature of the real task. Not a course about it — the task. Write the report, take the photos, build the tiny model, run the small event.
3. 03Spend half a day near it. Volunteer, shadow, help out, sit in. Proximity teaches you things nobody writes down.

Then compare the three by feel, not by prestige. Which one did you keep thinking about afterwards? Which one made you curious about the next layer down? Boredom in an experiment is a genuinely useful result: it eliminates a direction for a few hours' cost.

Aim for a direction, not a destination

You are not looking for the answer. You are looking for the next reasonable step, taken with slightly better information than the last one. Careers assemble themselves this way in practice, even the ones that look planned in hindsight. The person who seems to have known all along usually just kept moving and narrated it afterwards.

The skill worth building isn't picking correctly. It's getting good at exploring, testing, and updating — because the thing you pick will change anyway, and that skill is what carries you through when it does.