Technical Assessment: How to Design, Score, and Run One

A technical assessment is a task or test that shows whether a candidate can do the hands-on work a technical role needs.

For a recruiter, the assessment is a stage where strong candidates can drop out or a weak hire can slip through. A well-built assessment tests only what the job needs and gives the hiring manager or client evidence they can compare across candidates.

Start with your must-have skills before you pick a test

A technical assessment should test the skills the person can't do the job without, and nothing else. That sounds obvious, but the hiring manager or client may arrive with a test they used last time or one a colleague liked. So before you agree to any format, get the must-haves written down. If the request is still vague, the advice on writing the job description helps you pull the must-haves out of it.

Say you're filling a data analyst role for a regional hospital network, working with the analytics manager (or the client, if you're agency-side). Together you agree on the must-haves. The person needs to write queries in SQL (the language used to query databases), clean messy data, and explain results to managers who don't work with data. Those skills become the yardstick for every later choice.

The format, the time the task asks of candidates, and the scoring guide should each trace back to that list. If you want the research on how different assessment types compare, the guide to talent assessment types covers it.

Choose a technical assessment format that fits the role

The right format shows the must-have skill in action while asking for as little candidate time as possible. A technical skills assessment can take several shapes, and each one reveals something different. The table compares six formats by what each one shows, how much candidate time it asks for, and where it can mislead you.

For developer roles, a coding assessment often defaults to a live session or a take-home. A code review, where the candidate reads and debugs existing code, can show much of the same skill in less time. It's worth suggesting when the hiring manager or client worries about candidate drop-off.

Work samples deserve a closer look because they come nearest to the job itself. The U.S. Office of Personnel Management (OPM) assessment guidance for federal hiring describes them as tasks that mirror what employees do on the job. It adds that applicants often see work samples as very fair, but they can be costly to develop and need people to observe or rate the results.

That trade-off still points to a work sample for the hospital role. The candidate gets a small, made-up dataset, cleans it, writes queries to answer a handful of questions, and writes a short summary for a manager. Each part tests one must-have, and nothing in it needs real patient data or access to the hospital's systems.

Check the task design before you send it

A fair technical assessment follows a few design rules that you agree with the hiring manager or client before the first invite goes out. These are recommended practices rather than legal requirements. They matter because they keep results comparable from one candidate to the next.

• Tie each part to a must-have: If a part of the task doesn't test a must-have, cut it. Extra parts add candidate time without adding evidence anyone will use.
• State the expected time up front: Put it in the invite and keep it reasonable for the level of the role. A candidate with a full-time job may not have a free weekend, so an oversized task can push that person out.
• Use one task and a scoring guide written in advance: Before anyone sees a submission, describe what strong, acceptable, and weak work looks like. The interview scorecard guide shows a format you can adapt for written work.
• Have reviewers score separately before they discuss: When reviewers talk first, the most confident voice can end up setting the score. Scoring alone and then comparing shows where they disagree and why.

In the hospital search, the analytics manager's first draft was a long take-home that added a dashboard build and a statistics section. Neither was on the must-have list, so you push back. The task gets cut to the cleaning, the queries, and the written summary, and the invite states the expected time. Now every part of the candidate's effort goes to a skill the role needs.

Decide what outside help and AI tools are allowed

Decide which outside help is allowed before you send the technical assessment, then say so in the invite. Candidates may reasonably assume documentation, search, and AI assistants are fair game unless told otherwise, since people use those tools at work. A clear rule protects honest candidates from guessing wrong.

There's no single right rule, so pick one that fits the role and design the task around it. Some teams allow every tool and judge the result plus the reasoning behind it. Others ban AI assistants in a live session but allow them for a take-home. If AI help is allowed, a recall-style technical assessment test tells you little, so lean on tasks that need judgment about a specific dataset or codebase.

Using an AI tool isn't cheating unless the rules say not to. A fast submission or an unusual answer is a reason to ask questions, not proof of anything.

Whatever the rule, a short follow-up conversation is a practical check on the work. Ask the candidate to walk through their choices and make a small change on the spot. Someone who did the work can explain why they filtered or joined the data a certain way, whatever tools they used. The guide to problem-solving interview questions has prompts that work well here.

For the hospital role, the follow-up has the candidate walk the analytics manager through the results and then adjust one query to answer a new question. That conversation also tests the third must-have, because explaining results to someone who doesn't work with data is part of the job.

Run the recruiter's side of the assessment

The recruiter keeps the technical assessment moving by briefing the candidate, handling accommodation requests, chasing scores, and passing on feedback. Each of those steps shapes whether the candidate sees the process as fair. The brief comes first, and the invite should answer what the candidate would otherwise have to ask:

• Format: What the task is and how it will be delivered.
• Time: The expected time and the deadline for returning it.
• Tools allowed: The outside-help rule you set with the hiring manager or client.
• Who reviews it: The people or roles who will score the work.
• When they'll hear back: A date you can keep.

The same invite is the right place to ask about accommodations. The Job Accommodation Network's November 2022 guide covers hiring under the Americans with Disabilities Act (ADA). It says an employer may tell applicants what the process involves, such as a timed written test. It also says the employer may ask whether they'll need a reasonable accommodation, and that the applicant starts the process by requesting one.

That's the legal frame. As an operating habit, route any request to HR at once so the candidate isn't left waiting. For other legal questions, such as adverse impact (when a test screens out one group at a higher rate), the pre-employment assessments guide covers the basics. HR or employment counsel should make the final call.

Once tasks come in, chase the scoring. A submission that sits unreviewed tells the candidate the effort didn't matter, and a strong person may accept another offer meanwhile. The candidate experience plan has more on keeping each stage moving.

Last, pass on feedback. Someone who gave you their time should hear a specific reason for the outcome, tied to the scoring guide. The interview feedback templates help you word it.

Keep the technical assessment in proportion

A technical assessment is one signal, so pair it with a structured technical interview instead of letting the score decide alone. A 2022 re-analysis of earlier meta-analyses found that many selection methods predict job performance less well than older estimates said. Structured interviews came out as the top-ranked method in that review.

That finding doesn't make the test useless. It means the score belongs next to the interview evidence rather than above it. For the hospital role, the work sample, the follow-up conversation, and a structured interview each give the analytics manager a separate piece of evidence to weigh.

Use the assessment as evidence before you make the call

A good technical assessment stays small, clear, and close to the job. Candidates know what's expected and what help is fair. Reviewers agree on what good work looks like before they read a submission, and a short conversation afterward confirms the work and shows how the person thinks. Before your next technical search, sit down with the hiring manager or client and strike every part of the draft task that doesn't test a must-have.