Does Your Résumé Match the Job? One Posting, Three Résumés
A keyword match and an evidence match are not the same thing. A keyword match asks whether a word appears on your résumé. An evidence match asks which specific line of your history proves you can do the thing. Screening software is good at the first and mostly blind to the second — and the gap between them is where genuinely qualified people get filtered out, and where genuinely unqualified people get through to an interview that goes badly.
The video above shows the gap. One real job description, three résumés, five requirements, and a map from each requirement to the line that proves it — or to a blank.
The posting: what it requires, and what it only prefers
Morgan Stanley posted an Investment Banking Real Estate Associate / Vice President role in New York on 19 March 2026. It has since closed to applications, and the quotes below come from the syndicated copy of the posting rather than Morgan Stanley’s own careers site, which renders its listings client-side. It is a good worked example because it is unusually blunt about which way its gate swings:
- Required: “4-6+ years of relevant experience within the investment banking industry, or an MBA with 1-4+ years of experience, with a proven track record working on complex executed deals across M&A and various industry sectors.”
- Preferred, explicitly not required: “Experience in the real estate industry is strongly preferred, but not required.”
- Plus: strong written and verbal communication, commercial instinct under deadline pressure, ability to lead a team of analysts and associates, and Series licensing held or obtained after hire.
So the hard gate is investment banking execution. Real estate — the thing the job title is named after — is the soft preference. Hold onto that, because it is what makes the middle résumé interesting.
Why each résumé landed where it did
The first one: keyword match and evidence match agree
Marcus is a real estate investment banking associate. Every requirement has a line under it, and each line is a specific transaction with a number attached: an executed $1.4B portfolio sale, a $410M follow-on he drafted the S-11 for, three analysts across four live mandates.
Nothing clever is happening here. When someone has actually done the job, the keywords and the evidence point at the same thing, and any screening method — human or machine — reaches the same answer.
The second one: keyword-rich, requirement-blocked
Elena is the case worth studying. She has eight years in commercial real estate: $780M of loans originated, roughly $4.5B a year valued in ARGUS, NOI and DSCR and debt yield on sixty-plus assets annually.
Run her résumé against this posting as a bag of keywords and she looks strong. Real estate: yes, everywhere. Valuation, underwriting, modelling, ARGUS, the asset classes: all present, all genuine, all more prominent on her résumé than on a generalist banker’s.
Run it against the requirements and three of five rows come back empty. She has never executed a sell-side M&A process, never taken a company through public capital markets, and has not run a deal team. She has the thing the posting says it prefers and not the thing it says it requires.
That inversion is the entire point. The keyword scan and the requirement scan disagree — and the requirement scan is right. Any tool that ranks her highly because the words line up is setting her up for a first-round conversation that ends four questions in.
The third one: the control
Ray is a restaurant general manager: a 140-seat room, $6.8M in revenue, full P&L, 42 people, food cost taken from 34% to 28.5%. He is good at his job. He is also not an investment banker, and four of five rows come back blank.
The one row that fills is real: the posting asks for commercial instinct under deadline pressure, and renegotiating fourteen vendor contracts into a six-point margin improvement is evidence of exactly that. It is quoted from his own résumé, not narrated on his behalf. The other four stay empty because nothing in his history speaks to them.
What the screening software actually does
Automated screening is close to universal at the top of the funnel. In Harvard Business School and Accenture’s Hidden Workers: Untapped Talent — a survey of more than 2,250 executives across the US, UK and Germany — “more than 90% of employers in our survey use their RMS to initially filter or rank potential middle-skills (94%) and high-skills (92%) candidates.”
The same study is unusually candid about what that costs. 88% of employers agreed that qualified high-skills candidates are vetted out of the process because they do not match the exact criteria established by the job description. For middle-skills roles the figure was 94%. That is not job seekers complaining; that is employers describing their own funnel.
The report also names the mechanisms, and they are blunter than most people expect. Almost half of the companies surveyed automatically screened out any résumé showing an employment gap of more than six months — “the resume is automatically screened out by their RMS or ATS, based on that consideration alone.” And on credentials: “Such requirements winnow out applicants who have the right experience, but not the exact credential stipulated.”
Notice what all of these have in common. A gap is a date subtraction. A credential is a string. An exact criterion is a match. Every one of these filters operates on something countable, because countable is what software is good at.
One statistic to stop repeating
You will read everywhere that “75% of résumés are rejected by an ATS before a human ever sees them.” It is usually attributed to the Hidden Workers report. It is not in that report. The only 75% in it refers to something else entirely — the share of US employers using these technologies at all. We could not trace the 75% rejection claim to any primary research.
It matters because the myth teaches the wrong lesson. It tells you to beat a machine by stuffing keywords. The actual finding is worse and more useful: the filters are real, they are crude, and they cut on exact criteria — so the fix is to make sure your genuine qualifications are stated in terms a literal reader can match, not to manufacture qualifications you do not have.
What an experienced recruiter does with the same résumé
Then a human reads it, fast. Ladders’ 2018 eye-tracking study put the average initial screen at 7.4 seconds, and found that “recruiters in our study spent more time focusing on job titles than on any other element.” Treat that number with some caution — Ladders is a job board with a commercial interest in the finding, and its published methodology is three sentences long and discloses no sample size. The direction is well corroborated even if the decimal is not: the first pass is a skim, and titles carry it.
What survives the skim is a different question. NACE’s Job Outlook 2025 survey (237 employer respondents, fielded August–September 2024) found that nearly 90% of employers look for evidence of problem-solving ability on a résumé and nearly 80% look for teamwork skills, with written communication, initiative, work ethic and technical skills each named by at least 70%.
Read that carefully: employers are looking for evidence of those things. Not a claim of them. And they say the hard part is exactly that verification — in Employ’s 2025 Recruiter Nation Report (surveyed by Zogby Analytics in September 2025 across more than 1,200 US recruiters and HR managers), the most-cited obstacle to skills-based hiring was that skills are hard to measure and validate consistently — ahead of hiring managers’ preference for traditional credentials (46%) and the time required to implement it (46%).
So the two screens fail in opposite directions:
| The software | The recruiter | |
|---|---|---|
| Reads for | Exact strings, dates, credentials | Titles first, then evidence |
| Handles well | Hard skills, tools, certifications | Judgement, scope, whether a claim is credible |
| Fails by | Rejecting the right person for wording | Never seeing the résumé the filter cut |
| What moves it | Precision | Specificity |
Both are satisfied by the same underlying thing: a specific, checkable claim. “Strong stakeholder management” is invisible to a keyword filter and worthless to a recruiter. “Presented 24 credit memos to the bank’s investment committee on a two-week turnaround, 21 approved” is a phrase a filter can match and a fact a recruiter can interrogate.
The soft-skill problem, stated plainly
Every job description asks for communication, leadership and commercial judgement. Nearly every résumé claims them. Neither the software nor the recruiter can do anything with the claim, because a claimed soft skill costs nothing to write.
What both can use is a soft skill with a situation, a scope and an outcome attached. Compare:
- “Excellent leadership skills” — unverifiable, unmatched, ignored.
- “Led a team of three analysts across four live mandates, owning the CIM and management presentation on each” — a scope (three, four), a deliverable (CIM), and something an interviewer can ask three follow-up questions about.
That second version is what fills a row in the map above. The first one leaves it blank, and it should.
What this means for your résumé
- Map before you write. Take the five to seven things the posting actually requires and, for each, name the line in your history that proves it. Where you cannot, you have found the question you will be asked.
- Separate required from preferred. Elena’s problem was not a weak résumé, it was reading the title instead of the requirement. The word in the job title is often the preference; the gate is usually further down.
- Attach a number to every soft skill. Scope, duration, headcount, or outcome. Any of them turns a claim into evidence.
- Do not fill a gap with a synonym. A requirement you cannot evidence is not a wording problem, and rewording it only moves the failure from the résumé screen to the interview, where it costs more.
- Know when the honest answer is no. If the posting requires four years of a thing you have not done, name it early and ask what the realistic path is. Our guide to questions to ask in an interview has more on turning that into a useful conversation.
Our guide to preparing for an interview covers the rest of the method, and researching a company before an interview covers the half of prep that has nothing to do with your résumé.
FAQ
What is the difference between a keyword match and an evidence match?
A keyword match checks whether a word from the job description appears on your résumé. An evidence match checks whether a specific line of your experience proves you can do what the requirement describes. They usually agree at the extremes and disagree in the middle — someone can be keyword-rich for a role and still miss the requirement that actually gates it, which is the case the video above walks through.
What does an ATS actually scan for?
Countable things: exact strings from the job description, dates, titles, credentials and certifications. Harvard Business School and Accenture’s Hidden Workers survey found more than 90% of employers use their systems to filter or rank candidates initially, and named specific filters including a hard cut on employment gaps of more than six months and on exact credential matches.
Do applicant tracking systems really reject 75% of résumés automatically?
There is no primary research behind that figure. It is commonly attributed to the Hidden Workers report, which does not contain it. What that report does say is that 88% of employers believe qualified high-skills candidates get vetted out for not matching a job description’s exact criteria — a related but different, and better-sourced, problem.
How do I put soft skills on a résumé so they count?
Attach a situation, a scope and an outcome. NACE’s Job Outlook 2025 survey found nearly 90% of employers look for evidence of problem-solving and nearly 80% for teamwork — evidence, not assertion. “Strong communicator” does nothing; “presented 24 credit memos to the investment committee on a two-week turnaround, 21 approved” is matchable by software and checkable by a person.
How long does a recruiter spend on a résumé?
Ladders’ 2018 eye-tracking study reported an average initial screen of 7.4 seconds, with job titles drawing more attention than any other element. That study is vendor research and discloses no sample size, so treat the precise number loosely — but the underlying point, that the first pass is a fast skim anchored on titles, is well supported.
Bloom’s Interview Hub builds this map for you: it takes the job you saved and the résumé you already have, and lists each requirement next to the line from your own history that proves it. Where nothing backs a requirement, it says so instead of writing something flattering — because that blank is the question you are about to be asked.
Bloom is not affiliated with, endorsed by, or sponsored by Morgan Stanley. The posting is quoted as a publicly available example; the three résumés are fictional.