A career change is the goal people carry longest without starting. It has no deadline, no external pressure, and every reason to start next quarter is completely reasonable.
It is also the goal where coaching machinery earns its keep most obviously, because the failure is almost never a shortage of information.
Here is the structure, the questions worth running early, and an honest account of which parts software helps with and which parts it does not.
It is two goals wearing one label
Working out what to move to, and actually moving. They need different work and most people run them together.
Run together they interfere. You research options while half-heartedly updating a profile, make progress on neither, and after eight months conclude you are not serious enough. You were serious. The plan was two plans.
- Goal one is a research and elimination problem. It ends when you can name one direction in a sentence, and it needs a deadline or it never ends.
- Goal two is a logistics and nerve problem. It ends when you have moved, and it needs a runway number and a dated first action.
Doing goal one properly takes about six weeks. Doing goal two takes as long as it takes, and it cannot start until goal one has an answer.
Worth knowing: the reason goal one drifts is that research is indistinguishable from progress from the inside and produces nothing checkable. That is the same mechanism as procrastinating on the thing you actually want, and the tell is identical.
Weeks one and two: check the reason
A meaningful share of career changes are a response to one manager, one project or one bad year. That is a job change, and it is much cheaper.
Run these before you spend six months on the wrong problem:
- If your manager left tomorrow and the team changed, would you still want out? The single most useful question here.
- Which specific parts of the last year drained you? The commute, the meetings, the subject matter, the people, the pace, the lack of authority. Name the components rather than the job.
- Which parts did you like? There are always some, and people mid-frustration cannot list them without being asked.
- When did you last feel like this? If the answer is at your previous job, at roughly the same point, the pattern is worth understanding before you repeat it.
- What is the story you are telling about why? Write it down. Read it back in a week. Sentences that sounded true frequently do not survive being reread.
If the honest read is that it is this job rather than this field, stop here. You have saved yourself a year and the fix is a search rather than a change.
Weeks three to six: eliminate, do not explore
Exploring has no end state. Elimination does.
The structural trap is that reading about options feels productive forever. Set an end date at the start and work backwards from it.
- Write down every direction, including the embarrassing ones. Twelve to fifteen. The good one is rarely in the first three.
- Cut on constraints first, not on appeal. Money floor, geography, hours, retraining you are not going to do. Constraints are facts and appeal is a mood.
- For each survivor, name the actual daily work. Not the title. What you would be doing between nine and six. Most directions die honestly at this step.
- Rank the survivors by what you would have to give up, not by what you would gain. People overweight the gain and then discover the cost in month three.
- Pick one, and write it as a sentence with a date on it. "By March I am working in X" rather than "I am exploring X".
This is the stage a coaching session is genuinely good at, human or otherwise, because the work is producing options before judging them and then being made to commit to one. That is exactly what the GROW structure does, and the options stage has an explicit rule against evaluating early.
Weeks four onward, in parallel: talk to fifteen people
This is the step that changes the answer most and the one no software does for you.
Fifteen conversations with people doing the thing, in the actual job, at roughly the level you would enter at.
- Ask what the bad days look like, not what they love. Everyone can answer the first question honestly and nobody can answer the second one usefully.
- Ask what they did not know before they started.
- Ask how they got in, specifically, including who introduced them to whom.
- Ask what they would do if they were entering now.
- Do not ask for a job. It changes the conversation and you need the information more than you need the favor at this stage.
Fifteen sounds like a lot and it is about two a week for two months. It is also the single highest-return activity in the whole process, and it is the one that gets skipped, because it is the only step that involves being slightly uncomfortable in front of another person.
Weeks seven to twelve: the exit is a checklist
Not a leap. A list with numbers on it.
- Runway, as a number of months. Calculate it, do not estimate it. The number changes the decision.
- The floor. The lowest offer you would accept, decided now while you are calm rather than in month five while you are not.
- The first dated action. One, this week, with a day. Not "start applying".
- The skill gap, if there is one, with the cheapest possible test of whether you can close it. A short course or a small project beats a qualification you have not started.
- What you tell people, in one sentence. You say it two hundred times and the version you improvise the first time tends to stick.
Each of those is checkable, which is what makes this stage manageable in a way the research stage is not.
Go further: before the exit, run a pre-mortem. It is six months from now and the change did not happen: what went wrong? The twenty-minute version produces three or four if-then lines, and on a career change the answers cluster hard around money, one bad month, and a conversation you did not have.
Where software helps, and where it does not
Split honestly, because this is the section a vendor usually skips.
It helps with:
- The elimination structure. Fifteen options, cut on constraints, ranked by cost. That is a process and processes are what this is good at.
- The eleven o'clock problem. Career doubt does not arrive during office hours.
- Being the same across months. Friends get bored of your career change in week six, entirely reasonably. A model does not, and by month four that matters.
- Remembering what you said in week two, which is when your reasons were clearest and least contaminated by how the search has been going.
- The between-sessions half. Something arriving on a Thursday asking about the two conversations you said you would have is the mechanism the whole plan runs on.
It does not help with:
- The fifteen conversations. Nothing substitutes for them.
- Judging your read on people. Whether your boss is the problem, whether that offer is a trap. It has your account and your account is what needs checking.
- The nerve. Nobody outsources that.
- Anything that needs a network. Introductions come from people.
If you want the flat version of that boundary, what an AI life coach is and cannot do sets out the category without the sales pitch, and the comparison against a person is worth reading given a career change is exactly the kind of decision with other people in it.
Protecting the hours while still employed
The most common failure is not nerve. It is that the search never gets a slot.
A career change while working is roughly four to six hours a week: two conversations, some writing, some applications. Almost nobody has that in one block.
- One defended hour, same day each week, for the writing and the thinking.
- Conversations go in lunch hours and early evenings, because they need the other person's calendar rather than yours.
- Applications go in a fixed weekly slot, not whenever one appears, or the search runs your week.
- A named fallback for the week it all collapses, because it will.
Finding an hour that survives a bad month is the general version, and a career change is the goal that needs it most, because it competes directly with the job funding it.
The twelve weeks, in five lines
- Weeks 1 to 2: check whether it is the field or the job
- Weeks 3 to 6: eliminate to one direction, with a date on it
- Weeks 4 onward: fifteen conversations with people doing the work
- Weeks 7 to 12: runway, floor, first dated action, skill gap, your sentence
- Throughout: one defended hour a week, and a named fallback for the bad weeks
Common questions
How long does a career change actually take?
The deciding half is about six weeks if it has a deadline and indefinite if it does not. The moving half runs from three months to two years depending on whether you are changing function, industry, or both. Changing both at once is the version that takes longest and the one people underestimate.
Should I get a career coach instead?
If you can afford one, a specialist career coach is better than a generalist at the market-specific parts: what your experience is worth, how the hiring works, what your account of yourself sounds like from the other side. The generic machinery in this article is what a good one runs anyway.
Can an AI coach help me decide what to do with my life?
It can run a genuinely useful structured session on it, and that is worth having. What it cannot do is know anything about you that you did not tell it, which on a question this large is a real limit. Use it for the structure and the eleven o'clock conversations, and use people for the reality check.
What if I do not know what I want at all?
Then goal one is the whole project and goal two does not exist yet. Start with the elimination list rather than with introspection, because most people find it far easier to rule things out than to conjure a direction, and the ruling out is what surfaces the pattern.
Is it too late to change careers?
The question behind it is usually about cost rather than age, and the useful version is: how many working years are left, and what does staying cost across them. That is arithmetic and it is worth doing on paper rather than carrying around. The wider version of the feeling, stuck with nothing actually wrong, has its own page.
How do I stop the research phase running forever?
Put an end date on it before it starts, and make the deliverable a sentence rather than a feeling of clarity. Research with no output condition is a hobby that feels like work, and it is the single most common way this goal dies.
The takeaway
The reason career changes stall is almost never information. It is that two different projects are being run as one, neither has a deadline, and the only step that would settle the question is the one that involves talking to strangers.
Split them. Put six weeks and an end date on the deciding half. Have the fifteen conversations. Then treat the exit as a checklist with numbers on it rather than as a leap of nerve, because a checklist is something you can do on an ordinary Tuesday and a leap is something you postpone.
One thing to do today. Answer one question in writing: if your manager left tomorrow and the team changed, would you still want out? Ten minutes, and it decides whether the next six weeks are about a field or about a job.
