O*NET Interest Profiler Review (2026): What Its Own Reliability Data Shows

Most reviews of the O*NET Interest Profiler tell you it is free, credible, and published by the U.S. Department of Labor, then stop. That is all true and it skips the interesting part, which is that this is one of the few career instruments that publishes its own reliability data and invites you to check it. Almost nothing else in the category does, and the numbers turn out to say something more specific than "credible" about what the tool can settle for you.

I opened it this week and read the technical report behind it, so this review covers both: what the tool now does screen by screen, and what its published psychometrics license you to conclude. I build a career assessment myself, which makes me a competitor and also means I know which questions are worth asking about an instrument. The short version is that the Interest Profiler measures what it measures unusually well, and the reason it still leaves most people stuck is not a flaw in the measurement.

What you get, screen by screen

Start at onetinterestprofiler.org. If you have a link to mynextmove.org/explore/ip, that address now issues a 301 redirect to the new domain, so older guides still route you to the right place through an old door.

The web version runs 30 items. It says so before you begin: you will read 30 work activity descriptions and decide how much you would like doing each, on a five-point scale running Strongly Dislike, Dislike, Unsure, Like, Strongly Like. This matters for expectation-setting because the O*NET Resource Center also documents a 60-item Short Form, and that one is listed as paper-and-pencil only. If a guide tells you the online test is 60 questions, it is describing a version you will not be taking.

Two instructions on the intro screen are worth noticing, because they tell you what the instrument is deliberately excluding. It asks you not to think about the education a job needs or the money it pays, promising that comes later. And it says, in as many words, that this is not a test and there are no right or wrong answers. Both are the correct framing for an interest inventory: you are being asked what appeals to you with cost stripped out, which is a cleaner elicitation than asking what you want when the price is visible.

There is no account requirement. The tool offers to save progress on your device and states that no personal information such as your name or location is stored. An optional profile exists if you want results to persist. For a free government tool this is about as low-friction as it gets, and it is a genuine contrast with free career tests that collect an email before showing you anything.

At the end you get scores across the six RIASEC interest areas, and then the step the intro promised: you choose a Job Zone, which is how much preparation you are willing to invest, and the tool filters occupation matches accordingly. That second step is the part most reviews skip, and it is what turns an interest score into a list you can act on.

The reliability numbers nobody quotes

Here is what separates this instrument from nearly everything it competes with. The Department of Labor published O*NET Interest Profiler: Reliability, Validity, and Self-Scoring, a technical report with the actual coefficients in it. I read it rather than citing it by name, because naming a report is not the same as reporting what it says.

Internal consistency, which is how coherently the items on each scale hang together, measured on a sample where scale Ns ran from 1,024 to 1,045:

Interest scale	Coefficient alpha
Realistic	.93
Investigative	.94
Artistic	.94
Social	.95
Enterprising	.93
Conventional	.96

The report's own summary: the scales "had very high internal consistency estimates across all the RIASEC scales," with estimates "ranging from .93 to .96."

Stability, measured by giving the same instrument twice to 132 participants with roughly one month in between:

Interest scale	Test-retest correlation
Realistic	.87
Investigative	.81
Artistic	.88
Social	.92
Enterprising	.88
Conventional	.91

Those are strong figures. For comparison within the same report, the Interest-Finder, the instrument it was benchmarked against, retested between .79 and .88. The report also checked for response bias, the tendency to agree or disagree with everything regardless of content, and found the Interest Profiler's general factor accounted for 43.6% of variance against a benchmark of roughly 41% that Prediger established across 24 RIASEC datasets. In plain terms it is no more contaminated by yes-saying than the established instruments in its family.

What those numbers license, and what they do not

This is where most coverage of the tool goes wrong in a way that flatters it.

A test-retest correlation of .92 on the Social scale means that if you take this twice a month apart, you will get close to the same answer. That is a genuine property of the instrument and a good one. It is evidence the instrument is not noise.

It is not evidence that your interests are fixed for a career, and the study design cannot support that reading. One month is a short window. People do not usually change what appeals to them in four weeks, so a high correlation across that gap is close to the minimum you would demand of a working instrument rather than a discovery about human nature. Treating a one-month stability figure as proof that your interest profile will hold at 45 is reading something into the number that the design never tested.

High reliability tells you the instrument is measuring something consistently. It says nothing about whether that something is the thing your decision turns on.

And that is the sharper limitation. Reliability is about consistency, not about whether the measured variable predicts the outcome you care about. An interest inventory can be flawlessly consistent and still be silent on the question you brought to it, because what it holds constant is precisely the part of you that was already stable. It is not designed to detect what changed in your life this year, and it does not claim to be.

Where the career matches come from

The matching step is honest and worth understanding, because it explains both the strength and the ceiling.

Your six scores produce a profile. The tool then matches that profile against occupations in the O*NET database, which tags roughly 900 occupations with their own interest profiles, and filters by the Job Zone you selected. So the list you receive is "documented occupations whose interest profile resembles yours, at the preparation level you said you would accept."

That is a defensible piece of engineering, and notice what it is not. It is not a prediction that you will be good at those jobs, satisfied in them, or able to get hired into them. It is a similarity match against a catalogue. The catalogue happens to be the best public one in existence, which is why this beats the many quizzes matching you against thirty hand-written job descriptions, but a similarity match is still a similarity match.

The three things it will not tell you

Being specific about the gaps is more useful than a general "it is only a starting point."

Your situation. The instrument reads preference, not position. An undecided 19-year-old and a 41-year-old with a mortgage and eight years of accumulated expertise can produce an identical three-letter code, and their next moves have almost nothing in common. Nothing in the 30 items asks which of those two you are.

Whether the interesting thing will sustain you. Interest is measured at the level of "would you like doing this activity." Plenty of work is interesting to contemplate and depleting to do daily, and the reverse is also common. The gap between those two shows up months into a job, not on a five-point scale.

What it costs you. The intro screen explicitly removes education and pay from consideration, which is right for measuring interest cleanly and leaves the tradeoff entirely to you afterwards. The Job Zone step puts preparation back in, but nothing puts back the question of what you are willing to give up, which is usually where a career decision gets settled.

That last gap is the one I have the most data on. In our March 2026 survey of 439 people who completed the MyPassionAI career quiz, a self-selected group who came looking for career direction and skewing young at 77.7% aged 18 to 24, 70.8% said they want meaning and income together rather than one at the expense of the other, and only 6.4% were willing to defer passion and focus on income first. An instrument that brackets money out by design is measuring one axis of a decision that most people are making on two.

Who should take it, and what to pair it with

Take the Interest Profiler if you want a well-validated, free, no-account reading of which broad kinds of work appeal to you, especially if you are early and facing a field so wide that any credible narrowing is progress. On that job it outperforms most things you could pay for. If you want the framework behind it explained properly, including how to read your three-letter code, I wrote a longer piece on the RIASEC test and Holland Codes, and it sits in context against the other instruments in the best career aptitude tests ranked by what each one measures.

Pair it with something that reads the variable it brackets out. Interest tells you the flavour of work that appeals; it does not tell you which decision you are in the middle of. The MyPassionAI career quiz is built around that second variable: 25 branching questions, about three minutes, and it asks which struggle you are currently in and what you are optimising for over the next six months before it looks at anything else, returning one of 20 archetypes rather than an interest code. Where the Interest Profiler is deliberately stable, that one is deliberately sensitive to what changed.

Neither replaces the other, and I would rather be plain about the asymmetry than pretend otherwise: on measuring interest, the government instrument is better validated than mine and free, and it publishes evidence I cannot match. What it does not do is decide anything.

Where this leaves the tool

The O*NET Interest Profiler is the strongest free career instrument available, and unusually, you do not have to take that on trust. Internal consistency between .93 and .96, one-month stability between .81 and .92, response bias in line with the established RIASEC inventories, all published by the body that built it. Thirty items, five minutes of honest answering, no account, no upsell, matched against the best occupational database in existence.

Read the result as what it is. You have a reliable measurement of which activities appeal to you and a filtered list of documented occupations that share that profile. That is a map of the terrain. It is not a reading of where you are standing on it, and the reason so many people finish a well-built interest test still stuck is that they came with the second question and received an excellent answer to the first.

If the part you are missing is the second question, take the free career quiz now. Three minutes, and it starts from the situation you are deciding from rather than the preferences you have held since you were twenty.

Written by Marco Kohns, founder of MyPassion.ai, former Growth Product Manager at a venture-backed consumer startup operating in 100+ countries, ex-Techstars Berlin consultant, author of a Journal of Business Research paper on generative AI for growth hacking (MSc NOVA IMS Lisbon, 18/20).