What are your salary expectations? How to answer with a number

Give one number, say what it is anchored to, then stop talking. "For HR generalist roles in manufacturing I'm targeting $79,000" is a complete answer, and it beats a range, because a range gets heard as its bottom.

The sentence is the easy half. Choosing the number is the hard half, and the usual advice sends you to a national occupation median that may describe neither the job you applied for nor the industry paying for it. Two titles work as the test case, data analyst and HR generalist, and the national median fails them in opposite directions: too high for one, too low for the other.

What to actually say

Here is the whole script.

"I've been looking at HR generalist roles in manufacturing, and I'm targeting $79,000. I'd want to hear more about the scope before I go firmer than that - is that inside the band for this req?"

Three things are doing work in there. The number is specific, so it reads as researched rather than hopeful. The anchor is named, which means that to disagree, the recruiter has to disagree with an industry benchmark rather than with you. And the question at the end hands the turn back, which is the only way to learn whether you have just underbid without conceding anything to find out.

Which number goes in the sentence

Use the median for the industry you are applying into, not the median for the occupation. The Bureau of Labor Statistics publishes both, and they sit much further apart than most salary advice admits.

Human resources specialists - the occupation O*NET files "HR Generalist" under, code 13-1071.00 - had a median annual wage of $75,940 in May 2025. Split that same code by industry and it pays $84,100 in government and $60,550 in employment services. A $23,550 gap at the median, for work BLS classifies as one occupation.

Median annual wage for human resources specialists by industry, May 2025

One occupation code, five industries. The occupation-wide median was $75,940.

Government		$84,100
Professional, scientific, and technical services		$82,800
Manufacturing		$79,270
Healthcare and social assistance		$64,130
Employment services		$60,550

So an HR generalist who quotes the occupation median of $75,940 to a staffing agency has asked for roughly a quarter above that industry's typical pay, and will hear that the budget does not reach. The same figure quoted to a manufacturer sits a few thousand under the $79,270 manufacturing median - survivable, easily recovered later in the process. Quoted to a government employer, it hands back $8,160 before anybody has described the job.

The same split runs through the other example title, and it runs the other way. Data scientists had an occupation-wide median of $120,230 in May 2025, but $142,240 in publishing, broadcasting and content providers, $132,380 in computer systems design, $129,490 in credit intermediation, $128,050 at management-of-companies employers, and $108,650 at insurance carriers. Four of those five industries pay above the occupation median. A data analyst who quotes $120,230 to a publisher or a broadcaster is quoting a figure $22,010 below what that industry typically pays the code, and has done it while believing they aimed high - which is the failure mode nobody warns you about, because the national median feels like the safe, defensible answer and in four of the five industries BLS breaks out it is a discount.

Which is why "know your worth" is empty advice and "know your industry's median" is useful advice. Your worth is not a number that exists anywhere. The band the employer is working from is a number that exists, and it was written down before you applied.

What the spread tells you, and what it does not

The second figure worth having is the width of your occupation's pay distribution, because that width is how much room the question has in it.

Data scientists: the pay distribution, May 2025

$67,240Lowest 10 percent earned less than

$120,230Median annual wage

$199,130Highest 10 percent earned more than

BLS puts the data scientist median at $120,230 a year, or $57.80 an hour, as of May 2025, with the bottom tenth under $67,240 and the top tenth over $199,130. That is close to a threefold spread inside a single occupation code. In a spread that wide an unanchored number is nearly meaningless - $95,000 and $165,000 are both defensible, and nobody on the call can tell which end of that you think you belong at unless you tell them which end you think you belong at.

Here is the limit of this data, and it is a real one. BLS publishes percentiles, not experience levels. Nothing in the $67,240-to-$199,130 range tells you what a data analyst with three years earns against one with nine, or what the step up is for managing people, because the wage survey is not collected that way. Read the 90th percentile as "this occupation contains jobs that pay that," never as "this is what senior looks like."

BLS has no occupation called "Data Analyst"

That single fact wrecks more salary answers than any negotiating mistake does.

The nearest published series is Data Scientists, 15-2051, and its May 2025 median of $120,230 covers modeling work that plenty of people titled "data analyst" never touch. O*NET splits that code into specializations, cross-referencing Business Intelligence Analysts (15-2051.01) and Clinical Data Managers (15-2051.02). The reported job titles listed under Business Intelligence Analysts are Business Analyst, BI Analyst, Market Intelligence Analyst and their variants - much closer to what a typical data analyst posting actually describes.

So read the posting before you pick a series. If it asks for Python, statistics and model deployment, 15-2051 is genuinely your code and $120,230 is the right neighborhood. If it asks for SQL, Tableau and stakeholder reporting, that same figure is built partly out of machine-learning salaries, and quoting it means quoting somebody else's job. There is nothing dishonest about aiming high. There is a cost to aiming high with a source that does not support you, because a recruiter who works this occupation every day knows the difference and now knows you did not check.

The two example titles, side by side

The wage rows are all May 2025 Occupational Outlook Handbook figures; the last two rows are the 2025–35 projection round.

Because both columns come from the same survey and the same projection round, they can be read against each other. They are national, and they are not adjusted for where you live - which is the second thing this table does not tell you.

Do you have to name a number first?

If the posting carries a range, no. That range is your anchor and BLS becomes the sanity check: quote a figure in the upper half of the posted range when your experience meets or beats what the posting asks for, and use the industry median only to judge whether the whole band is low.

If the posting carries no range, somebody has to go first, and it is usually you. The question exists to avoid spending four interview rounds on a candidate who is outside budget, which is why it turns up at the phone screen rather than at the offer. You may decline to answer. Declining is also expensive, because a recruiter working through a screening list rarely chases the one candidate who would not say.

On the legal side, briefly. A number of states now require pay ranges in postings and restrict questions about salary history, but those rules change often enough that any count printed in a blog post is stale within a quarter, so I am not giving you one to memorize - check your own state labor department page before the call. If you are asked what you currently earn and would rather not answer, redirecting to expectations is usually accepted: "I'd rather anchor on the role - I'm targeting $79,000 for this one."

Who is actually deciding whether your number fits

Not the recruiter asking it. At most companies past a certain size the range was set before the requisition opened, by a compensation analyst working from survey data. BLS counts 118,100 compensation, benefits, and job analysis specialists in 2025, with a median wage of $78,210 as of May 2025, and building and maintaining the bands your answer gets checked against is the job.

Those specialists are spread nearly as wide as the people they set pay for: under $49,480 at the tenth percentile and over $128,920 at the ninetieth, again in May 2025. Which is a small, useful reminder that "the compensation team" is not a single authority with one view of the market. It is a job with juniors in it.

Credentials are a weaker lever here than most candidates expect. For human resources specialists, 47 percent of O*NET respondents said a bachelor's degree is required for new hires; for data scientists it is 48 percent bachelor's and a further 44 percent master's. When almost everyone in the applicant pool clears the same educational bar, the degree is a gate you already passed, not an argument for the top of the band. Scope is the argument - headcount supported, systems owned, revenue touched.

That changes what a good answer sounds like. You are not persuading anyone in that moment; you are being sorted into "inside the band," "top of the band," or "out." Nobody reopens a band because a candidate was eloquent about it. What does move is landing in the top half of an existing band with a reason attached, because the recruiter then has something concrete to write down when they go and ask for the high end.

It is also the mechanical reason a range costs you. Say "$72,000 to $86,000" and you have told a person accountable for closing the role inside a budget that $72,000 closes it. Say "$79,000, which is about the manufacturing median for this title," and the conversation starts at $79,000 with a citation attached. For scale, BLS puts the median annual wage across all occupations at $50,980 in May 2025 - so both of those answers are well above typical pay, which is precisely why an employer will not treat either as self-evidently reasonable, and why you need the anchor rather than the confidence.

What to do this week

Find your occupation code before you find your number. Search the Occupational Outlook Handbook for the work you actually do rather than the title on your badge, and where two codes plausibly fit - as they do for most data analysts - pick the one whose task list matches the posting in front of you.

Write down two figures: your occupation's national median, and the median for the industry the employer sits in. The second one goes in the sentence. If the industry table on the OOH page does not list your employer's industry, use the occupation median and say so out loud, because a named source you have slightly misapplied still beats a number from nowhere.

Then say the sentence once, aloud, before the call. It is one clause and a dollar figure, and the only way it goes wrong is hesitating in the middle of it, which reads as a number you do not believe yourself.

Keep that figure consistent everywhere it gets asked, because it gets asked more than once - the screen, the desired-salary box on the form, the offer call. HeroApply fills the application form from your profile, that field included, so the number you settled on is the number that goes in.