A CV is a document of claims: titles held, institutions attended, years survived. A portfolio is a document of evidence: the thing itself, inspectable, attributable, judgeable by anyone competent in the craft. That the first remains the default currency of hiring while the second is treated as a nice extra is one of the stranger equilibria in management practice, because the research has been pointing the other way for decades.
The classic reference is Schmidt and Hunter's 1998 synthesis in Psychological Bulletin, which compiled 85 years of personnel-selection research across 19 methods. Work sample tests, where candidates perform a piece of the actual job, showed a mean validity of .54 for predicting job performance, among the strongest single predictors reviewed, and combining general mental ability with a work sample reached .63. The details of such estimates get revised as methods improve, but the ordering has proven stubborn: samples of the work outperform stories about the work. Education, the credential line the CV leads with, sat far down the list.
None of this should surprise anyone. The best predictor of whether someone can do a thing is watching them do the thing. The puzzle is not the finding; the puzzle is the industry that read it and kept sorting PDFs.
What each document can and cannot lie about
Consider what a CV can misrepresent: scope ('led' a project one attended), causality (revenue that grew for reasons unrelated to the bearer), and context (a title that means different things in different firms). Verification requires calls, and the calls go to referees the candidate selected. The CV is not worthless; it is testimony, and testimony has its uses. But it is testimony delivered by the interested party, formatted by convention to remove exactly the texture an evaluator needs.
A portfolio can also mislead: work can be inflated, collaborative work presented as solo, and polish can be purchased. But its failure modes are inspectable. An evaluator can probe a portfolio in conversation ('walk me through this decision; what did you try first; what broke?') and watch the depth of the answers. Faked ownership collapses under three follow-up questions. A faked title survives every question, because there is nothing beneath it to probe.
The asymmetry extends to what each rewards over a career. A CV rewards accumulation of positions; a portfolio rewards accumulation of finished, showable work. People manage what their evidence system measures. A profession that hires on portfolios gets practitioners who finish things.
The objections, taken seriously
Portfolio hiring has real costs, and pretending otherwise discredits the argument. Reviewing work takes expert time; scanning a CV takes seconds, and at a thousand applicants the seconds win. Some work is invisible or confidential: the operations manager's portfolio is the absence of disasters, and much of it belongs to a former employer. And portfolios can encode privilege, since building public work requires slack that caregivers and the twice-shifted rarely have.
Each objection has an engineering answer. Screening cost: use small, paid, standardized work samples instead of demanding pre-existing bodies of work, which also levels the privilege problem by letting the evidence be produced inside the process rather than before it. Confidential work: structured debriefs of past decisions function as verbal portfolios, and their validity logic is the same, sample the thinking, not the biography. What has no answer is the status quo's implicit claim that a list of nouns predicts performance better than observed work. It does not, and the field has known it since before most current HR software existed.
The quiet reallocation of trust
Beneath the tooling question sits a question about where trust should live. Credential-based hiring places trust in institutions: the university, the previous employer, the title-granting hierarchy. Evidence-based hiring places trust in observable competence and in the evaluator's own judgment. The first scales cheaply and fails silently. The second costs more per decision and fails loudly, which is the failure mode organizations actually learn from.
Employers do not have to choose purely. But the default should invert: evidence first, biography as background. Read the work, then the story. Any hiring process that does it the other way around is admitting it finds stories easier to read than work, which is true, and is exactly the problem.
Sources
- Schmidt & Hunter (1998). The Validity and Utility of Selection Methods in Personnel Psychology (Psychological Bulletin 124, 262-274). Across 85 years of research and 19 methods, work sample tests showed mean validity of .54 for job performance; GMA plus work sample reached .63. https://home.ubalt.edu/tmitch/645/session%204/Schmidt%20&%20Oh%20validity%20and%20util%20100%20yrs%20of%20research%20Wk%20PPR%202016.pdf
- Burning Glass Institute & Harvard Business School (2024). Skills-Based Hiring: The Long Road from Pronouncements to Practice. Removing credential proxies without installing evidence-based evaluation changed fewer than 1 in 700 hires. https://www.burningglassinstitute.org/research/skills-based-hiring-2024