For Half of New-Economy Roles, the Resume Is the Wrong Document

The resume isn't dying. It's being demoted. For engineering, design, writing, and data work, it's now the cover sheet: the page that gets you to the gate, while a portfolio is the document that gets you through it. A resume is an unverified claim. A repo, a shipped feature, a published byline is proof. The market is repricing claims down and proof up.

Hiring managers in design and engineering will tell you this themselves. They decide off the GitHub or the case study, then read the resume to confirm what they already saw. The resume routes them to the evidence. The evidence makes the call. Most job seekers have this exactly backwards: they pour their prep into polishing the cover sheet and build almost nothing underneath it.

Why was the resume always a weak signal?

Because it predicts job performance badly, and it always has. When industrial psychologists measure how well common hiring screens predict actual on-the-job performance, the resume's two headline ingredients come out near the bottom. Job experience sits around r=0.07 and years of education around r=0.10, among the least valid predictors in the entire toolkit. Translate that out of correlation-speak: the things a resume mostly lists explain a rounding error of why one hire works out and another doesn't.

Now compare the thing a portfolio actually is: a work sample. The earlier line of research put work-sample tests at r=0.54, far ahead of experience or education. Later work was more conservative. Roth, Bobko, and McFarland's 2005 meta-analysis revised work samples down to r=0.33, still multiples better than what the resume leans on.

The kicker: 70.1% of US employers still analyze resumes despite this being well-documented for decades. The convention outlived the evidence.

What makes a portfolio harder to fake than a resume?

A resume is a self-attestation. A portfolio is auditable by anyone with a browser. That gap is the whole game.

The resume's trust problem is getting worse, not better. Employment-verification discrepancies jumped 44% between 2021 and 2024, from 9.9% to 14.26%, and 46% of all reference and credential checks now surface a mismatch between what a candidate claimed and what actually happened. The same research found 80.4% of Gen Z candidates admitted to lying on a resume, well ahead of every older cohort.

Read those two numbers together. Four in five of the youngest cohort admit to padding the page, and nearly half of all checks now surface a mismatch. The resume's signal-to-noise is degrading from both ends: candidates inflate, and the verification net catches only a slice of it. Public work is the structural opposite. It's verified by the fact that it exists. You can't fake 400 commits, a deployed app with real weekly users, or a newsletter archive a manager can read in ninety seconds. The proof is the artifact.

Where is this already true?

In the fields where the work is inspectable, the decision already happens off the work. Two candidates, same role, every time:

Software engineering. Candidate A's resume says "3 years Python experience," no links. Candidate B has a GitHub repo with 400 commits, a deployed side project with 200 weekly users, and a README that explains the architecture trade-offs. Any process that actually opens the evidence picks B by a margin the resume alone would never reveal.

UX design. Candidate A has a master's in HCI and a resume full of Fortune 500 logos. Candidate B has no degree, a Figma case study of a checkout redesign with an A/B result of +18% conversion, and a Dribbble following. Senior UX directors say it plainly: the portfolio is king, work experience matters, the degree is the least important of the three.

Data science. Candidate A has a CS degree from a mid-tier school. Candidate B has no degree, a Kaggle Grandmaster rank in the global top 0.1%, and three notebooks showing the methodology. The Kaggle rank is a work-sample test in disguise, and it's the one that moves.

Writing and content. A writer with 8,000 newsletter subscribers and 60 public essays applies for a content role. The resume says "freelance writer, 2022 to present." The manager googles the byline first, and by the time they glance back at the resume the decision is made.

Role	Cover sheet (resume)	The real file (portfolio)
Software engineer	"3 yrs Python"	GitHub repo, deployed app, architecture README
UX designer	Degree + brand logos	Figma case study with measured outcome
Data scientist	CS degree	Kaggle rank, notebooks with methodology
Content / writer	"Freelance, 2022 to present"	Public archive, subscriber count, bylines

The pattern doesn't change between fields. The proof carries the signal the resume only gestures at.

Notice what each strong candidate did. They didn't write a better claim. They produced a thing a stranger could open and judge. That's the move the resume can't make for you, because the resume's whole format is testimony. The work is exhibit.

Doesn't a strong resume still get you the interview?

Often, yes, and that's exactly the trap. The resume's job is to survive the first filter. Getting through that filter and getting picked are two different contests, and they reward two different documents.

Think of it as a relay. The resume runs the first leg: it gets you past the keyword scan and the recruiter's six-second skim, into the pile of people a hiring manager will actually consider. Then the baton passes. The second leg is the part that decides the offer, and on that leg the manager is looking for evidence they can trust. A resume that wins the first leg and has nothing behind it loses the second leg quietly, and you never find out why.

This is why polishing the resume past "clear and honest" has steep diminishing returns. Once it routes correctly, more wordsmithing buys you almost nothing. The same hour spent shipping one more commit, one more case study, one more published piece compounds into the document that wins the leg that actually matters. The resume is a gate. The portfolio is the room.

If skills matter so much, why is the shift so slow?

Because what companies say and what they do are two different data sets. This is the part you have to be honest about, or you'll over-rotate.

The rhetoric is real and rising. 85% of employers reported using skills-based hiring in 2025, up from 81% in 2024, while only 67% still use resumes as a screening tool, down from 73%. Job posts are loosening too: LinkedIn postings without a four-year-degree requirement rose from 15% to 20% year over year, a 33% relative jump.

The behavior lags hard. Tech job posts dropping degree requirements grew 240% faster than posts that kept them, but actual hiring of non-degree candidates in tech grew only 3% faster. Companies changed the listing, not the hire. A Harvard Business School and Burning Glass study put a sharper edge on it: degree-dropping moved fewer than 1 in 700 hires to a non-degree candidate, and only 37% of companies that dropped the requirement meaningfully changed who they hired.

So the transition is real but uneven. The degree still works as a lazy proxy wherever no portfolio exists to replace it. The candidate who burns down their whole approach to go portfolio-first is, in many pipelines, optimizing for a future that hasn't fully arrived.

What does this approach cost, and who does it punish?

Building proof favors people who already have time and access. A GitHub history, a polished design portfolio, a 60-essay archive: those take uncompensated hours, decent hardware, and usually a day job flexible enough to allow side projects. Candidates from lower-income backgrounds or with caregiving loads are structurally disadvantaged here, and pretending otherwise is dishonest.

The cold-start problem is real too. The portfolio answer assumes you've already done work worth showing, which is a catch-22 for anyone in their first professional role.

And the proof signal itself is starting to erode. The same way keyword-stuffing killed the ATS-scanned resume, AI-generated code, AI-written articles, and AI-designed mockups are flooding portfolio platforms. Managers are already getting more skeptical of unverified portfolio claims. Proof beats assertion, but only while the proof is still costly to fake.

Some fields are credential-locked by necessity, not laziness. Medicine, law, civil engineering, regulated finance: the license is the gate because liability demands it. The portfolio argument simply doesn't apply there. "Half of new-economy roles" means knowledge work where the output is inspectable, not the whole labor market.

What should you actually do now?

Don't abandon the resume. Reframe its job. Treat the resume as the table of contents and the portfolio as the chapters.

1. Pick the one artifact your field actually inspects. Engineer: a repo with a real README. Designer: one case study with a measured outcome. Writer: a public archive with a count attached. Data: a notebook or a ranked submission. One strong piece beats five thin ones.
2. Make it auditable. Deploy the thing. Show the number. Link it. A claim a manager can verify in ninety seconds is worth more than a paragraph they have to trust.
3. Cut the resume down to a router. Its only job now is to be honest, crisp, and to point cleanly at the proof. Every line should survive a manager who's already seen your work.
4. If you're early-career, manufacture the work sample. No track record means you build a small, real, finished thing on purpose. The catch-22 breaks the moment you ship something that exists.

This is consistency over intensity in its plainest form. One shipped repo with steady commits beats a heroic week of resume rewording, because the repo compounds and the resume just sits there. The deep want underneath all of it is to be evaluated on what you can actually do. Proof is how you force that evaluation instead of hoping for it.

Want to figure out which document is doing the real work in your field, and build the proof that gets you through the gate? Message Praxy on WhatsApp. We'll audit your cover sheet and map the one artifact worth building next.