How to Research What a Job Pays Before You Apply

At Metaintro, we watch listed compensation move across 50 million jobs every day, and the single most common mistake we see job seekers make is anchoring on one stray number. The fix is not a paid tool or a hunch. It is a short, repeatable research workflow built on free primary data, starting with the government agency whose entire job is measuring what Americans earn: the U.S. Bureau of Labor Statistics. This guide is one piece of our larger pillar on understanding what a job really pays, and it focuses on the research step, where the real numbers come from and how to combine them into a range you can actually defend.

Why does the first salary number you find usually mislead you?

Type "average salary" for almost any role and the first result is an average. An average is a single number that gets dragged upward by a handful of very high earners, which means it usually overstates what a typical person in that job makes. It also hides the thing you most need to see, which is the range. Two people can hold the same title and earn amounts that are $50,000 apart, and the average tells you nothing about where you would land between them.

The bigger problem is provenance. Most of the "salary" numbers floating around the open web are self-reported, aggregated from an unknown mix of people, with no way to check the year, the location, or how many responses fed the figure. When you cannot answer "who reported this, from where, and when," you cannot defend the number in a conversation, and you certainly cannot budget your life around it.

Neutral government data solves all three problems at once. The Occupational Employment and Wage Statistics program surveys roughly a million establishments over a three-year cycle and reports wages for specific, coded occupations. It gives you percentiles instead of a lonely average, it tells you exactly which month and year the data represents, and it breaks results down by metro area. That is the difference between a number you found and a number you can stand behind. For a deeper walk through building the number that fits your exact situation, see our guide on how to figure out your market rate.

How do you find your occupation code in BLS data?

The Bureau of Labor Statistics does not organize pay data by job title, because titles are messy. Your "growth marketer" and someone else's "demand-gen lead" might be the same work under different names. Instead, every occupation gets a Standard Occupational Classification code, a six-digit number that pins your role to a defined set of duties. Software developers are 15-1252. Registered nurses are 29-1141. Customer service representatives are 43-4051. Find the code that matches what you actually do, not the fanciest title on the org chart.

To find yours, go to the OEWS tables, search by keyword or browse the occupation groups, and read the short description to confirm the duties line up with your day-to-day. This one step is where most casual research falls apart. People search a trendy title, get a number for a slightly different job, and anchor on it. Matching the code first is what makes every figure after it trustworthy.

Here is the before and after. Before: a job seeker Googles "average salary for a developer," takes the first number, and walks into interviews quoting a figure that came from an unknown blend of senior and junior roles in unknown cities. After: the same person looks up software developers, SOC 15-1252, reads the actual percentile spread, and knows that the national median is $135,980 as of the May 2025 OEWS release, with the 25th percentile at $105,210 and the 75th at $171,980. One of these people is guessing. The other is quoting a source they can name.

What do the OEWS percentiles actually tell you?

The reason percentiles matter is that they replace a single dot with a full picture of the pay distribution. For software developers in the May 2025 OEWS data, the spread runs from $82,460 at the 10th percentile to $214,670 at the 90th, with the median sitting at $135,980. That is a $130,000 gap between the bottom and top of the same occupation. If you only saw the $148,100 mean, you would miss that most developers earn well below it.

Read the percentiles as a map of where you might sit based on your experience and skills. The 25th percentile ($105,210 for that role) roughly marks early-career or lower-cost-of-living territory. The median ($135,980) is the middle of the market. The 75th percentile ($171,980) is where senior, specialized, or high-demand-metro talent tends to land. Your job is to locate yourself honestly on that ladder, not to claim the 90th percentile because it is the biggest number on the page.

This works for any occupation, not just high-paying ones. Customer service representatives, SOC 43-4051, show a May 2025 median of $44,770, with the 10th percentile at $31,750 and the 90th at $63,590. Registered nurses, SOC 29-1141, post a median of $97,550. The pattern holds across the whole economy, so the same three-minute lookup gives you a defensible middle and a defensible ceiling for almost any role you are targeting. If you want a sense of what "good" means relative to these spreads, our guide on what counts as good pay puts the percentiles in context.

Why does the metro number matter more than the national one?

A national median is an average across the entire country, which means it blends together a high-cost coastal metro and a low-cost rural county into one figure that describes neither. The moment you apply for a specific job in a specific place, the national number becomes background context and the metro number becomes the one that matters.

The gap is not small. For software developers, the national median in the May 2025 OEWS data is $135,980. In the Austin-Round Rock-San Marcos metro area it is $134,120, right around the national line. In the San Francisco-Oakland-Fremont metro area it is $186,640, more than $50,000 higher for the very same occupation code. If you quoted the national median in a San Francisco interview, you would be leaving real money on the table without knowing it. If you quoted a San Francisco figure for an Austin role, you would price yourself out.

So always pull the metro-level percentile for the place the job is based, using the same OEWS tables filtered to your metropolitan area. Where the role is fully remote, look at the metros where the employer concentrates hiring and treat the range as a band rather than a point. Cost of living is the reason these numbers diverge, and it is also why the same salary feels generous in one city and tight in another, a tension we unpack in what annual income means for your budget.

How do you tell whether the data is current enough to trust?

Old pay data is quietly dangerous, because compensation moves every quarter and a two-year-old figure can be off by real money. The OEWS release is labeled with the exact reference period, and the current tables reflect May 2025 estimates, so you always know how fresh your anchor is. That labeling is itself a reason to prefer government data. A random web result rarely tells you its vintage at all.

To adjust a slightly older figure or simply to understand which direction pay is heading, use the Employment Cost Index, which the Bureau of Labor Statistics publishes quarterly. For the 12-month period ending in March 2026, total compensation costs rose 3.4 percent and wages and salaries rose 3.4 percent, according to the ECI. That tells you a median measured a year ago is probably a few percent light today, and it gives you a defensible, sourced way to say so rather than pulling a growth number out of thin air.

The broader wage backdrop helps too. The U.S. Census Bureau reported median household income of $83,730 for 2024 in its Income in the United States: 2024 report, a useful reality check on where any single salary sits relative to a typical household. Across the roles we track at Metaintro, we consistently see that seekers who understand the current direction of pay, not just a static figure, negotiate with far more confidence, because they can explain why last year's number is a floor rather than a ceiling.

How do pay-range posting laws hand you free primary data?

One of the most valuable pay signals now shows up right inside the job posting itself, for free, and most applicants scroll past it. A growing number of states require employers to include a good-faith salary range in their listings. New York, for example, spells out its posting requirements through the New York State Department of Labor, and states including California, Colorado, Illinois, Maryland, Minnesota, Washington, and others have their own versions on the books as of 2026.

These posted ranges are primary data in the purest sense. They are what a real employer, right now, has committed in writing to pay for a real role. When you see the same role posted with ranges by several employers in your metro, those ranges become a live, self-collected dataset that you can compare directly against the OEWS percentiles. If postings cluster near the OEWS median, you have strong confirmation. If they run high, you have evidence the market is hot for that role right now.

Read the ranges with a clear eye. Employers set them wide to preserve flexibility, and the top of a posted band is usually reserved for the most experienced candidates. Use the bottom of the range as a realistic floor and the OEWS median as your center of gravity. For the full picture of what these laws let you ask and how far transparency extends, see pay transparency laws and what you can ask, and to catch listings that fall suspiciously below the market, our guide on spotting underpaid roles and lowball offers. This same posted-range data is what powers a strong internal case too, which is why our guide on how to ask for a raise leans on the exact numbers you are collecting here.

How do you triangulate three sources into one defensible range?

No single source is the whole answer. The skill is triangulation, laying the government percentile, the pay-trend data, and live posted ranges next to each other and reading where they agree. When two or three independent sources point at the same band, you have a range you can state out loud without flinching.

Pay pressure tracks the same shift in hiring. Lacey Kaelani, CEO of Metaintro, told People Managing People that "AI is not completely eliminating roles, but instead restructuring roles and therefore slowing hiring for some jobs." Knowing the real number for a role is how you protect your pay when that restructuring reaches your field.

Here is the workflow in five steps. First, find your Standard Occupational Classification code in the OEWS tables so you are pricing the right job. Second, pull the metro-level percentiles for your area and note the 25th, 50th, and 75th figures. Third, check the Employment Cost Index to see which way pay has moved since that reference period and nudge your figure accordingly. Fourth, collect the posted ranges from several real listings in your metro that are covered by pay transparency rules. Fifth, set your defensible range from the overlap, typically from around the OEWS median up toward the 75th percentile, adjusted for how you compare on experience and skills.

Here is the second before and after. Before: asked about expectations, a candidate says "around $60k?" with a rising, uncertain inflection, and the number is a guess that invites a lowball. After: the same candidate says the OEWS median for this role in their metro is a specific figure, that postings they have tracked cluster right around it, and that given their experience they are targeting the upper half of that band. One answer is a shrug. The other is a sourced position that shifts the conversation. When you are ready to turn that range into a script, our guide on how to negotiate a job offer and negotiating salary in an interview show exactly how.

Where do skills and AI fit into your pay research?

Two roles can share a Standard Occupational Classification code and still pay very differently, because specific in-demand skills carry a premium that a broad occupation median cannot capture. This is where you layer skill data on top of the government baseline. The PwC 2026 Global AI Jobs Barometer found that workers with AI skills command a wage premium of roughly 62 percent, up from 57 percent the prior year, which is exactly the kind of within-occupation gap the median alone hides.

Demand for those skills is broad, not niche. Gallup reported that AI use at work reached 45 percent in the third quarter of 2025, up from 40 percent the quarter before, according to its AI Use at Work Rises analysis. So when you research what a job pays, ask not only "what does this occupation earn" but "where do I sit within it given my skills." If you carry a skill the market is paying up for, you have a defensible reason to target the upper end of the OEWS spread rather than the median.

Modern research tools can do this layering for you. Rather than reading percentile tables by hand, an AI-powered career platform can match your specific skills against live listed compensation and show where you fall inside the range. That is the practical spirit of our guide on using AI to benchmark your salary, which sits alongside our broader AI skills that keep you employable work and our AI-powered job search playbook. The wage split is documented in the full PwC 2026 Global AI Jobs Barometer findings. In the postings on our platform we see pay for the same title vary widely by skill and metro, which is the whole reason a research workflow beats a single number.

Why does base salary alone give you an incomplete answer?

Even a perfect base-salary range is only part of the picture, because total compensation includes elements that never appear in a headline number. Bonuses, equity, retirement match, health premiums, and paid time off can swing the real value of an offer by tens of thousands of dollars. The Employment Cost Index itself tracks benefits separately from wages precisely because they move independently and matter enormously.

So once your base-salary research is solid, extend the same rigor to the full package. Ask what a "3 percent match" or "unlimited PTO" is actually worth in dollars, and weigh a lower base with strong benefits against a higher base with thin ones. A rushed applicant compares two base numbers and picks the bigger one. A prepared one builds a total-compensation view and sometimes finds the "lower" offer is worth more. We break this down in compare total compensation, not just salary, and if you are weighing structure itself, hourly vs salary and gross vs net pay round out the money picture. Once an offer lands, knowing how to read a pay stub confirms the real dollars are showing up the way the research said they would, and if your target sits near a milestone, how much is six figures reframes what that number actually buys.

What number can you actually defend in a live conversation?

The point of all this research is not to sound smart. It is to reach a moment, when someone asks "what are you looking for," where you can answer without your voice going up at the end. So build toward one clear sentence you can say out loud.

Name your Standard Occupational Classification code so you know you are pricing the right job. State the metro-level median from the OEWS, because place changes everything. Note that posted ranges in your metro, under pay transparency laws, cluster near or above that median. Adjust for the direction of pay using the Employment Cost Index, and adjust for your own skills, which can carry a premium the median hides. Then land on a band, not a single point, and say where inside it you sit and why.

That sentence, "the median for this role in my metro is X, postings I track cluster near it, and given my experience I am targeting the upper half," is the whole payoff. It is calm, it is sourced, and it cannot be waved away, because every piece of it traces back to a neutral primary source you can name. When you build a number that way, you stop hoping for a fair offer and start explaining what a fair offer looks like. What is the one number you could defend right now if someone asked, and which of these three sources are you still missing to get there?

Related Articles

• Career Metrics, Earning Potential, and Market Value in 2026
• How to Negotiate Salary (Complete Guide)
• 15 Salary Negotiation Tactics That Actually Work
• Americans Demand Salary Transparency
• Could AI Actually Raise Your Salary?
• What Software Engineers Actually Earn in 2026 by Level and City
• How to Negotiate Salary After a Job Offer, Scripts That Win
• How Much Should Your Salary Increase When Changing Jobs
• Why a $180K Tech Salary No Longer Stretches in San Francisco
• Companies That Pay for Vacation in 2026 and the Benefits Worth Chasing

People Also Asked

Q: What is the best free source for salary data?

A: The Occupational Employment and Wage Statistics program from the U.S. Bureau of Labor Statistics is the strongest free source, because it surveys real employer payrolls and reports full percentile ranges by occupation and metro area. Start by finding your six-digit Standard Occupational Classification code in the OEWS tables, then read the 25th, 50th, and 75th percentiles for your specific metro rather than the national average.

Q: How do I know if a posted salary range is fair?

A: Compare the posted range against the OEWS percentiles for the same occupation code in the same metro. If the range brackets the OEWS median, it is a genuine market range. If the whole band sits below the 25th percentile, treat it as a lowball and dig into total compensation before deciding. Posted ranges exist because pay transparency laws now require them in many states, so collect several and look for the cluster.

Q: How current is BLS wage data?

A: The current OEWS tables reflect May 2025 estimates, and every release is clearly labeled with its reference period so you always know the vintage. To account for movement since then, the Employment Cost Index showed wages and salaries rising 3.4 percent over the 12 months ending March 2026, so a year-old figure is likely a few percent low today and should be treated as a floor.

Wondering what you should earn? Metaintro tracks how roles get posted and paid in real time, so you can benchmark your number against the live market instead of guessing. Sign up to see what your target role pays and get matched to openings that fit.