You want out. Maybe it's the job, maybe it's the whole field, maybe it's just Monday.
Quitting is terraform apply on your life. Let's run plan first.
How do you change careers? Work out whether the field is really the problem. Compare your role's skills with the target role's in real job postings. Close the gaps that show up, and test the move with a small project while you still have a salary. Job titles are a bad guide to how big the move is. The skill diff is a better one.
TL;DR
- Titles hide the distance. Product Manager and Product Owner share 9 of their top 10 skills. Marketing Manager and Data Engineer share none.
- Ignore the skills everyone lists. Three of them sit in the top 10 of most roles. Take them out and 81% of role pairs have nothing in common.
- Check the door before the room. 42% of Data Analyst postings are entry level, against 18% for Data Engineer.
- Deploy to a canary first. One small project in the new field teaches you more than ten articles. Including this one.
First: is it the field, or just this job?
No dataset here, only my opinion. A surprising number of "wrong career" bugs turn out to be config errors.
Ask yourself three things:
- Would a new manager fix it? Then you need a new manager.
- Would another company fix it? Same job, different logo. Much cheaper than a career change.
- Would two weeks off fix it? That was burnout. Go and rest, the backlog will still be there. It always is.
If you still dread the core of the work after all three, the problem may well be the field. Read on.
Titles are a lossy format
We track the skills employers ask for in 44 AI-related roles, which makes 946 possible pairs of roles. I compared the top 10 skills of every pair, because a diff is how I look at most things.
A few results:
- Site Reliability Engineer and Cloud Engineer: 10 of 10 shared. Two titles, one pager.
- Product Manager and Product Owner: 9 of 10. In the Netherlands, the advertised median for a Product Owner is €6,000 a month, against €5,500 for a Product Manager. That's 81 and 65 postings, so it may just be noise. Still, I'd read a few of those ads.
- Marketing Manager and Data Engineer: 0 of 10. Possible! But this is a rewrite, and nobody has ever finished one of those on schedule.
Now the catch. Prompt Engineering and AI Tools Literacy are each in the top 10 of 35 roles, and Data Analysis in 33. So plenty of roles look related when they only share those three.
Counted plainly, 18% of pairs share no skills at all. Remove those three and it jumps to 81%.
Every service in your company depends on the same logging library. That doesn't make them the same service. So when you compare two jobs, skip what everybody asks for and look at what's left.
Three sizes of move
Here's what the diff looks like in practice.
The patch: Account Manager to Account Executive. They share 6 of 10 skills, but the weights change. AI Sales Tools goes from 17% of postings to 63%, CRM Automation from 23% to 59%. You probably know the work already. Now show it on your CV.
The migration: Financial Analyst to Data Analyst. Also 6 of 10, and that number flatters it. Excel Automation falls from 83% to 13%. SQL, in 63% of Data Analyst postings, isn't in a Financial Analyst's top 10 at all. Your macros were a fine legacy system. They don't migrate.
The rewrite: anything at 0 of 10. Expect to learn a new core skill set and prove it from scratch. This is where a course or a licence can earn its cost. Check the requirements first, because regulated fields don't accept "I watched the videos".
Our career paths lay out this diff for two common moves, including Data Analyst to Data Scientist. Your target role may be moving too, so check how AI is changing jobs as well.
Check the door before the room
A small skill gap doesn't help if nobody hires newcomers. So look at the level mix too. Of postings with a level:
- Sales Development Representative: 82% entry level
- Data Analyst: 42%
- Data Engineer: 18%
- Legal Counsel: 14%
My read, and it's only that: if your target has a narrow door, go through a neighbour. Data Analyst and Data Engineer share 3 of their top 10, and the first takes far more beginners. It's a two-hop route, and two hops beat a connection refused.
Canary, then rollout
Nobody sane ships a big change to all of production on a Friday. (I have seen it done. I have also seen the Saturday.) Do the same with your career:
- Run a canary. Build one small thing the target job would pay for: a dashboard, a volunteer project, a task at your current job. You learn whether you like the work, and you get a story for the interview.
- Talk to someone who moved recently. They still remember what got them hired. Ask what they actually did last Tuesday.
- Set a timeout. Pick a date and a test: "By 1 March I've shipped the project and sent 10 applications." "Gradual" with no deadline is a process that hung in 2023 and is still showing as running.
- Size your runway yourself. I have no solid data on how long a switch takes or how far pay drops, so I won't make up a number. Write down your monthly costs, your training costs and how many months you can run below that floor.
Being new again will feel bad for a while. You were senior, and now you're asking what an acronym means. That's a cold cache: slow at first, and no sign that the hardware is broken.
What we built for this
Three parts of SlashHash help here. None of them can tell you whether to jump.
- AI Readiness Score. It maps your skills against current demand in the Netherlands, and shows adjacent domains your skills could move into. It's in beta.
- Discuss my fit. Puts your CV next to a job and lists strengths, gaps and what to prepare. Slash, our assistant, can also draft a short plan to close a gap.
- Tailor my CV. Suggests rewrites of your own sentences for that job. It asks about missing facts and never invents them. If your CV and the job barely overlap, it says so.
FAQ
How do I know if I should change careers? Test the cheaper fixes first: a new manager, another company, time off. If you'd still avoid the core work of the job after all of those, the field itself may be the problem.
Which career changes are easiest? Those with the largest overlap in specific skills. In our data, Product Manager and Product Owner share 9 of their top 10 skills, and Account Manager and Account Executive share 6. Ignore skills nearly every role lists, such as Prompt Engineering and Data Analysis.
Can I change careers without going back to school? Often, if the gap is a tool or a skill you can show with a project. Regulated fields are the exception: if the job legally needs a licence or degree, you need it too. Read the requirements before you pay for anything.
How long does a career change take? We don't have reliable data on that. The size of the skill diff and any required credentials will decide most of it, so set your own deadline and review it.
How do I explain a career change to a hiring manager? Say why this field and why now, then give one concrete example of doing the work: a project, a volunteer role or a task from your current job. Proof beats a polished story.
Your move this week
- Run the diff. Open your role page and the target role's page. Skip the skills everyone lists and write down what's missing.
- Read 10 real postings for the target role. Count how many you could apply to today, and how many are entry level.
- Pick one canary and one date. One small project, one deadline, in your calendar.
For the Dutch side of this (retraining, pay, language), read how to switch careers in the Netherlands. Thinking about paying for help? See what career coaching actually fixes.
Figures: SlashHash AI careers snapshot, September 2026: AI-related postings in the Netherlands and Australia over the previous 180 days. It covers AI-related postings only, so its skill lists lean toward AI and it may undercount general skills such as negotiation. Overlaps compare each role's 10 most requested skills, so a skill just outside the top 10 counts as absent. Entry-level shares are as classified from the postings. Salaries are advertised medians from postings that quote pay. The overlap counts and the three-sizes framing are my own analysis of that data.