June 24, 2026

Researching Job Outcomes Before Choosing Your Career Path

Person at a crossroads surrounded by floating career symbols and question marks

Only 25% of recent graduates describe themselves as being on their dream career path, according to ZipRecruiter's 2026 Annual Graduate Report. One in four. The other three ended up somewhere reasonable — a job that "makes sense," a stepping stone role, something fine enough on paper — without ever knowing whether it was actually the right choice.

That isn't a failure of ambition. It's a failure of research. And the gap between "ended up somewhere" and "chose deliberately" is almost entirely about what you do before committing.

The Problem With How Most People Decide

Standard career advice follows a predictable script: figure out what you're passionate about, match it to a career, look up the average salary, maybe shadow someone for a day.

Passion fails as an early filter. Research from 80,000 Hours, a career research nonprofit that has studied this for over a decade, finds that passion for work tends to develop after gaining competence — not before. Using it to filter at the research stage often eliminates careers you'd find deeply satisfying once you were actually good at them.

Personality tests don't save you either. Myers-Briggs and Holland Type assessments correlate with job performance at roughly r=0.24. That's barely better than chance. Career counselors don't lead with this caveat because these tools form the backbone of most career education programs.

And average salaries? They describe a population, not you. The median income for a field says nothing about distribution tails, geography, or trajectory — the three variables that determine your specific outcome.

The real issue is a category error. Most people research what a career looks like in theory. What they need is research into what outcomes look like for real people who already made that choice.

Start With the Government Data (Seriously)

Two public data sources cover most of the ground, and neither requires a subscription.

The BLS Occupational Outlook Handbook covers more than 800 occupations with 10-year employment projections, median wages, and entry requirements. Total U.S. employment is projected to grow by 5.2 million jobs between 2024 and 2034 — but that growth is sharply uneven. Healthcare, clean energy, and tech infrastructure dominate projected gains. Certain clerical and manufacturing occupations show flat or declining demand. The OOH makes this visible at the occupation level, not just the industry level.

The Federal Reserve Bank of New York's College Labor Market database is less famous but more useful for anyone choosing a degree. As of Q1 2026, the unemployment rate for recent college graduates sits at 5.7%. That headline sounds manageable. What gets less attention is the underemployment rate: 41.5%. Nearly half of recent graduates work jobs that don't require a college degree.

The NY Fed breaks this down by major. Computer science and nursing graduates show underemployment rates near 20%. History and psychology graduates often exceed 55%. For fields with chronically high underemployment, the writing was on the wall in this dataset for years — most people just weren't looking.

Most job seekers check average salary in a field. Fewer check whether graduates actually work in that field five years later. That gap is where bad decisions hide.

Reading Salary Numbers Without Getting Burned

Salary figures lie by omission. Not intentionally — they aggregate badly, and most salary guides don't explain why.

The first problem: medians hide distribution tails. A strong median software engineering salary doesn't show you the bottom quartile, which often earns roughly half the median and clusters in less technical roles or oversaturated markets.

Second, starting salary versus career trajectory. A teaching position that starts at $43,847 in a state with step increases, union protections, and a defined-benefit pension looks completely different at year 15 than at year 1. PayScale tracks how wages evolve over time by occupation and experience level. That growth curve is more useful than comparing entry-level pay across fields.

Third: geography shapes salary more than most people account for. A data analyst earns $110,000 or more in San Francisco and around $54,000 in Indianapolis. National averages blend those extremes into a number that describes almost nobody's real situation.

Tool	Best For	Key Limitation
BLS Occupational Outlook Handbook	10-year demand projections	National averages only
NY Fed College Labor Market	Underemployment and outcomes by major	Covers recent grads, not mid-career
PayScale	Salary by experience and location	Self-reported, over-indexes tech workers
O*NET OnLine	Task-level job descriptions	Less useful for salary research

Triangulate across at least two of these. A career that looks strong in one source but weak in another is worth deeper investigation — not a quick rubber-stamp.

What Numbers Miss: The Human Layer

Data tells you what typically happens. It can't tell you what the job feels like to do for 2,000 hours a year.

Informational interviews close this gap, and they're dramatically underused. Indiana State University's career center found that 54% of students who completed informational interviews reported tangible outcomes: referrals (19%), interview invitations or job offers (17%), and maintained professional contacts (18%). That's research that simultaneously builds your network.

The format is simple: a 30 to 60-minute conversation with someone currently doing the job. Not to ask for work. To understand what the work actually looks like.

Questions that generate genuine signal:

What does a genuinely bad week look like in this role?
Where do most people in this field actually top out — what's the realistic ceiling?
Who burns out and leaves within five years, and what were the warning signs?
If you were starting over with current information, would you make the same choice?

These questions surface what the BLS will never show you. They're also more reliable than "do you like your job?" (which almost always produces socially comfortable non-answers rather than useful ones).

Job shadowing adds observation to conversation. Watching someone work for a half day reveals things even honest practitioners forget to mention: the pacing, the interruptions, the emotional register, the team dynamics. Aim for three to five informational interviews minimum — below that, you're sampling too narrowly to spot patterns.

Testing the Hypothesis Before You Commit

80,000 Hours ranks research methods by their correlation with actual job performance. The findings are worth sitting with.

Work samples — actually doing tasks representative of the job — correlate with performance at r=0.33. Structured interviews with practitioners come in at r=0.42. Standard personality tests land at r=0.24.

Doing beats asking, and asking beats taking a quiz. In practice, this points toward a progression:

Read and observe (2-5 hours) — BLS data, LinkedIn profiles of people 10 years into the role, Reddit communities from practitioners
Interview practitioners (5-20 hours) — 3-5 structured conversations with people actively working in the field
Complete a short project (1-4 weeks) — freelance work, a volunteer engagement, or a course with substantive applied output
Formal trial (2-24 months) — an internship or apprenticeship before committing to a graduate program or full career pivot

Most people jump from step 1 directly to step 4. The gap between "I researched it online" and "I enrolled in a graduate program" is exactly where the expensive mistakes happen.

ZipRecruiter's 2026 data puts numbers on this. Only 5.3% of graduates completed apprenticeships, but 32.9% of those completers reported being on their dream career path with zero role misalignment. That's the highest satisfaction rate of any pathway in the survey — higher than traditional four-year programs, higher than internships alone. Short trials produce better data than long commitments. That's the research finding most people consistently act against.

The Experience Gap That Shapes Everything

Here's the finding that should lead every career research conversation but rarely does.

Work experience during college is the single strongest predictor of post-graduation employment, per ZipRecruiter's 2026 report. Among graduates with college work experience: 81.6% found employment. Among those without it: 40.7%. That 41-percentage-point gap exceeds the effect of major, school selectivity, or GPA in most hiring research.

This reframes what "researching your career" should actually mean. It isn't only reading and interviewing. It includes accumulating early experience in the field, even in small doses.

Internships are the obvious vehicle. But freelance work, part-time roles, and side projects with demonstrable output all count too. A student who spent a semester doing pro-bono digital marketing for a local business has better firsthand data on whether she wants a marketing career than someone who spent that semester reading about it (and yes, she's also more employable — the research pays off twice).

The 41.5% underemployment rate for recent graduates isn't random noise. It's largely a function of people entering fields without having validated their fit or built a track record. The research gap and the skills gap are the same gap. Both are closed by the same action: start doing relevant work before you need a job in the field.

Bottom Line

Use the right data sources first. The BLS Occupational Outlook Handbook and the Federal Reserve Bank of New York's College Labor Market database are free, authoritative, and most people have never opened them. Start there before looking at salary aggregators.
The underemployment rate matters more than median salary. A field with a 55% underemployment rate means most graduates aren't accessing those headline wages. Check the NY Fed breakdown by major before committing to a degree.
Talk to practitioners before you trust the data. Three to five informational interviews, structured around bad weeks and realistic ceilings, will tell you things no spreadsheet can.
Test with work before committing to programs. A four-week freelance project gives you better predictive data than any personality assessment. Work samples correlate with job performance at r=0.33; Myers-Briggs at r=0.24. Do the math.
Experience is research. Graduates with college work experience are employed at twice the rate of those without it. Start accumulating relevant experience early — it's both how you learn what you want and how you prove you can do it.

The 25% of graduates who say they're on their dream path didn't get lucky. They did the homework first.

Frequently Asked Questions

Is the BLS Occupational Outlook Handbook actually reliable for career decisions?

Yes, with caveats. The OOH is most reliable for established occupations with long hiring histories — healthcare, education, skilled trades, engineering. For fast-moving fields like AI/ML engineering or climate tech policy, supplement it with more recent data from LinkedIn Workforce Reports or industry associations, since BLS projections can lag the market by a few years. The trend direction is usually sound even when specific numbers are slightly dated.

How many informational interviews do I need before I can trust my conclusions?

Three to five is the practical range. Below three, you're sampling too narrowly — one unusually positive or negative experience can skew your entire read. Above five, you're usually hitting diminishing returns unless you're getting wildly inconsistent answers (which itself tells you something — it usually means outcomes vary heavily by employer, geography, or specialty). When multiple practitioners describe the same frustrations independently, trust the pattern.

Is it a myth that you should "follow your passion" when choosing a career?

It's not a myth exactly — passion matters — but as a decision filter it's applied at the wrong stage. The research from 80,000 Hours suggests passion follows competence more than it precedes it. People who love their work usually got there by getting good at something first, not by identifying a passion and chasing it. Using passion as your primary research criterion eliminates careers you'd find rewarding once you had real skills in them.

What should I do if the salary data for my target career looks discouraging?

Dig into the distribution before drawing conclusions. National medians mask significant geographic variation and experience-based growth. A role that pays $47,000 as a starting salary might reach $91,000 at the 10-year mark with specialization, or pay very differently across cities. PayScale lets you filter by years of experience and location. The median describes the middle of a population — not your specific trajectory in a specific city with a specific specialization.

How do I research outcomes for newer fields that don't show up in BLS data yet?

Work backwards from the skills. New fields often lack their own job titles in the OOH, but their underlying skill sets do appear elsewhere. If you're interested in AI policy, look at what think tanks, government agencies, and tech companies currently pay for policy analysts and what they list as requirements. LinkedIn job postings filtered to the last 90 days give you a real-time picture of what employers are actually asking for and paying — more current than any government database.

Is job shadowing worth the effort compared to just doing informational interviews?

They answer different questions, so they're not interchangeable. Informational interviews tell you what practitioners think about their work. Job shadowing shows you what the work actually looks like in motion — the pace, the interruptions, the moments of frustration that don't come up in a polished 45-minute conversation. If you have time for one, do the interview. If you can do both, the combination gives you a much more accurate picture than either alone.