
In the 2022 meta-analysis by Sackett and colleagues, structured interviews had the highest predictive validity among the selection methods studied (0.42), compared with 0.19 for unstructured interviews. The same principle applies when evaluating a freelancer’s portfolio: set criteria upfront, verify their individual contribution, check relevant references, and use a short, paid work sample.
At a glance
- Start with the brief: decide which deliverables the project needs before opening any portfolio.
- Always separate project outcomes from the freelancer’s individual contribution: ask who made which decisions.
- If work samples are limited or confidential, use anonymized case studies, private presentations, and a practical test.
- Compare candidates using the same scorecard, with 3–5 criteria and scores from 1 to 5 defined in advance.
- Pay for the test, clarify in writing who owns the rights to the work produced, and handle candidate data in accordance with the GDPR.
Contents
- How to review a portfolio for the role and project
- The criteria that matter: quality, consistency, results, and relevance
- When work samples are limited or confidential
- Structured interviews and references: how to verify what you see
- How to design a realistic, paid work sample
- Comparing candidates with a scorecard
- Privacy, intellectual property, and fair testing
- Our perspective
- Pre-vetted talent for your project
- Frequently asked questions
- Sources
How to review a portfolio for the role and project
A portfolio shouldn’t be judged in a vacuum: it should be compared against the work you need done. Before reviewing it, write three lines outlining what you need delivered, by when, and under what constraints. Then look for evidence of that kind of work in the portfolio.
Each role offers different kinds of evidence:
- Brand designer: complete identity systems (logo, color palette, typography, applications), not just standalone logos.
- Performance marketer: campaigns described in terms of channel, budget managed, objective, and key metric.
- Copywriter or content creator: published work, along with the original brief and the tone the client requested.
- Developer: live products, accessible repositories, and explanations of architectural decisions.
The trickiest part is identifying the individual contribution. Almost all professional work involves a team, and a successful project alone doesn’t tell you who made it successful. If a case study reports a 40% increase in sales, that number is the project’s outcome: the freelancer needs to explain which part they handled, which decisions they made, and which were made for them.
The law also recognizes that a work can have multiple legitimate authors. Article 10 of Italy’s copyright law establishes that when contributions are indistinguishable and inseparable, copyright belongs jointly to all coauthors. That’s why the right question isn’t “Is this your work?” but “Which part of this work is yours?” For each project, ask about their role, the people involved, the timeline, the tools used, and what they would do differently. Precise answers, with details only someone who actually did the work would know, are worth more than any polished layout.
The criteria that matter: quality, consistency, results, and relevance
Years of experience and project counts look impressive, but predict little. In Schmidt and Hunter’s 1998 meta-analysis, which brought together 85 years of selection research, years of work experience had a predictive validity of just 0.18. What matters more is the substance of the work and how closely it resembles your project.
Four criteria cover almost every situation:
- Quality of execution: attention to detail, error-free work, and command of the tools of the trade. A designer aligns elements, a copywriter catches typos, and a developer writes readable code.
- Consistency: the standard stays high across all work, not just the first item on the list. A portfolio with one excellent project and five mediocre ones suggests an inconsistent professional—or a project whose success depended mostly on the team.
- Results: clearly stated, measured objectives, with a defined timeframe. A figure without context (“+200% traffic”) always needs a baseline, a duration, and a channel.
- Relevance: an industry, client size, deliverable type, and constraints similar to yours. Someone who has only worked for major brands with large budgets may struggle at a startup that needs speed on limited resources, and vice versa.
Add a fifth element: reasoning. The best portfolios show the initial problem, the alternatives ruled out, and the rationale behind the final choice. That’s where you learn how someone will approach your project when conditions differ from the examples shown. Record each assessment alongside the relevant criterion: you’ll need those notes to compare candidates consistently.
When work samples are limited or confidential
Many skilled professionals have sparse portfolios for entirely legitimate reasons: they do white-label work for agencies, have signed nondisclosure agreements, or have just left an in-house role. In Europe, trade secrets are protected by EU Directive 2016/943, implemented in Italy through Legislative Decree 63 of 2018. A responsible freelancer can’t show you everything—and shouldn’t.
In fact, a candidate who sends you another client’s confidential material is showing you how they would handle your data. Look for other evidence instead:
- Anonymized case studies: industry, problem, approach, and aggregated results, without the client’s name or sensitive data.
- Private presentation: a screen-sharing session where the freelancer walks you through the work without sending files.
- Personal or pro bono projects: these often reveal style and approach more clearly because they’re created without external constraints.
- Redacted documents: editorial calendars, wireframes, or reports with names and figures concealed.
- Direct references: a former client can confirm the freelancer’s role and results even when the work isn’t public.
For early-career professionals, shift the emphasis from past experience to present performance. A well-designed work sample is the closest measure of actual work you can get. It puts someone with ten years of clients on equal footing with someone who has just one year but a sound approach. Just remember to assess potential using the same criteria you apply to everyone else, without lowering or raising the bar.
Structured interviews and references: how to verify what you see
The portfolio shows what was done; the interview checks how. The difference between a free-flowing interview and a structured one is measurable: in Sackett’s data, validity rises from 0.19 to 0.42—more than double. Structured means three things: the same questions for every candidate, questions tied to actual work, and answers assessed against a scale defined in advance.
The most useful questions start with the portfolio itself:
- “Choose the project you’re proudest of and tell me about the hardest decision you made.”
- “In this case study, what would you have done differently with twice the time, or half the budget?”
- “Tell me about a time a client changed direction halfway through a project. How did you handle it?”
- “How do you keep people updated on progress when you’re working remotely?”
References round out the picture, but use them thoughtfully: in Schmidt and Hunter’s meta-analysis, reference checks had a validity of 0.26—useful, but lower than work samples. Make them more reliable by asking specific questions rather than a generic “How was it working with them?” Ask two former clients what role the freelancer played, whether they met deadlines, how they responded to feedback, and whether they would hire them again for a project similar to yours.
For help setting up the rest of the relationship, from the first call to delivery, see our complete guide to working with a freelancer.
How to design a realistic, paid work sample
Among all the methods analyzed by Schmidt and Hunter, work samples had the highest predictive validity: 0.54. But a test only works if it resembles real work and respects the professional’s time.
Five steps to design one:
- Base the test on the actual project: outline a landing page, turn a brief into three concepts, or sketch out a campaign plan. Never use a generic task.
- Keep it short: a few hours, not days. If you need more, you’re asking for an assignment, not a test.
- Write complete instructions: objective, materials, constraints, delivery format, and evaluation criteria.
- Pay for the test: agree on the fee upfront. The freelancer issues a proper invoice, and those just starting out can get guidance on registering for an Italian VAT number as a freelancer.
- Assess the process too: questions asked, deadlines met, and clarity of the handoff.
Italian law is clear on payment. Article 3 of Law 81 of 2017 deems clauses setting payment terms beyond 60 days from receipt of the invoice abusive. Pay for the test promptly, just as you would for an assignment.
If the project involves specific tools, such as a CRM, the test should reflect that context. Before defining it, clarify what kind of professional you need—for example, by comparing options for implementing HubSpot with a partner, consultant, or freelancer. Delivery and payment terms should be in writing: our guide to freelance contracts covers the clauses you shouldn’t leave out.
Comparing candidates with a scorecard

Judging three different portfolios on gut feeling almost always rewards the best presenter, not the best professional. Structure makes a difference here too: in Schmidt and Hunter’s study, structured interviews reached a validity of 0.51, versus 0.38 for unstructured interviews. The same principle applies to portfolios and tests: decide on criteria and a scoring scale before reviewing candidates.
A simple scorecard works better than a complex one. Choose 3–5 criteria, assign each a weight, and describe what scores of 1, 3, and 5 mean:
| Criterion | Weight | 1 (below expectations) | 3 (meets expectations) | 5 (excellent) |
|---|---|---|---|---|
| Relevance to the project | 30% | No similar work | Similar work in industry or format | Similar work in industry, format, and constraints |
| Quality of execution | 25% | Obvious errors | Clean, with no surprises | Above-average attention to detail |
| Individual contribution | 20% | Vague role | Clear role in part of the work | Owned and explained key decisions |
| Work sample | 15% | Misses the brief | Meets the brief | Meets the brief and suggests well-reasoned improvements |
| Communication | 10% | Slow or unclear responses | Timely | Proactive, clarifies questions in advance |
Three rules make the scorecard reliable. First: two people assess candidates independently and compare scores only at the end. Second: if scores for the same criterion differ by more than two points, discuss them with your notes in hand. Third: no off-scorecard bonuses; if something matters, include it in the criteria for everyone. That way, weeks later, you’ll still be able to explain why you chose one freelancer over another.
Privacy, intellectual property, and fair testing
Evaluating a freelancer means processing personal data: resumes, contact details, references, and sometimes interview recordings. EU Regulation 2016/679 (GDPR) requires you to collect only necessary data (Article 5), inform people about the purpose and duration of retention (Article 13), and provides for fines of up to €20 million or 4% of annual worldwide revenue in the most serious cases (Article 83). In practice: provide a short privacy notice at the start of the selection process, set a deletion deadline for unsuccessful candidates’ data, and don’t record anything without explicit consent.
When it comes to work produced during the test, watch out for a point many businesses overlook. Article 4 of Italy’s Law 81 of 2017 establishes that the economic rights to original contributions created while performing a contract belong to the self-employed professional, unless inventive activity is the subject of the contract and is compensated as such. If you want to use the test output, put it in writing and pay for it; otherwise, don’t use it. For more on rights transfers and licensing, read our guide to copyright and freelancers.
Finally, fairness. Give every candidate the same test, the same time, the same information, and the same point of contact. Don’t request personal data that isn’t needed to evaluate the work, and send everyone feedback at the end, even if it’s brief. The freelance world is small: how you select people today determines who will respond to your next search.
Our perspective
Every day, we see businesses choose a freelancer for a beautiful portfolio, only to discover a month later that talent wasn’t the issue—fit was. Technical skills and cultural fit are two separate assessments, and you need both.
That’s why selection at Hooop isn’t an automated filter. We evaluate professionals rigorously and individually: we look at the work, but above all at the thinking behind it, their actual role in projects, and how they collaborate with others. A portfolio tells you what someone can do; a well-run conversation tells you how they’ll do it with your team.
There’s another factor businesses underestimate: a good freelancer doesn’t work in isolation. When a project requires multiple skill sets, how those professionals interact matters. That’s exactly why we built our internal ecosystem: to assemble cohesive teams with centralized management, so our clients can focus on strategy instead of coordinating individuals.
The method we describe in this guide is the one we use: clear criteria, concrete checks, and human relationships before standardized forms.
Marco Petrazzi, CEO of Hooop
Pre-vetted talent for your project
If you don’t have the time or team to build scorecards, interviews, and tests for every search, we can handle it for you. With Talent as a service, we introduce professionals already vetted for skills and fit; with Hooop Studio, we build a tailored team for marketing, design, branding, content, and development; with Fractional Management, you bring experienced leaders into your business for as long as you need them.

Explore our services on the for businesses page, or start with our free business assessment to identify the skills your project is missing. If you’re still defining the role, our guide to finding qualified freelancers will help you structure your search.
Frequently asked questions
How many projects should a good freelance portfolio include?
There’s no magic number: relevance and consistency matter more than quantity. A few projects similar to yours, clearly explained and with a well-defined individual role, are worth more than a long gallery of unrelated work. If samples are limited, supplement them with references and a work sample test.
Is it fair to ask for an unpaid test?
A test that produces usable work should be paid, and the fee should be agreed on beforehand. Remember that under Italy’s Law 81 of 2017, economic rights to original contributions remain with the freelancer unless otherwise agreed. A paid test also attracts in-demand professionals, who usually turn down unpaid ones.
How can I tell whether a portfolio result is really down to the freelancer?
Ask which decisions they made, who they worked with, and what constraints they were given. Then verify their account with a direct reference from the client or project team. Vague answers about a very specific result are a sign you should dig deeper.
How long does it take to properly evaluate a freelancer?
It depends on the role, but with a scorecard ready, reviewing the portfolio takes little time per candidate, followed by a structured interview and a test lasting a few hours. Preparing criteria and questions once saves time on every subsequent search.
Sources
- Sackett et al. (2022), Revisiting meta-analytic estimates of validity in personnel selection, Journal of Applied Psychology
- Schmidt and Hunter (1998), The validity and utility of selection methods in personnel psychology, Psychological Bulletin
- Italian Law No. 633 of April 22, 1941, copyright protection (Normattiva)
- Directive (EU) 2016/943 on the protection of trade secrets (EUR-Lex)
- Italian Law No. 81 of May 22, 2017, protections for non-entrepreneurial self-employment (Normattiva)
- Regulation (EU) 2016/679, GDPR (EUR-Lex)

