Work sample tests are proven to be one of the best predictors of success on the job, so what happens when candidates use ChatGPT to prepare for them? Here’s how we see it: AI is just another option in our productivity toolbelts. It shouldn’t matter which tools candidates use at any stage of hiring. Work sample testing helps you assess candidates’ skills in key areas and whether they understand the context of the job. With the right approach, whether a candidate uses AI (or any other resource) to do the work becomes irrelevant.
Identify Your Goal and Make Sure the Task Is Relevant to the Job
Think about the key skills needed to succeed in the role. What do you want to see in action? What do you want to learn from a candidate’s work? What will convince you that this person has the skills to do the job and collaborate with your team?
- Ensure that the work assignment closely aligns with the actual responsibilities of the job and the same 3-5 skills you are evaluating at every step of the hiring process. The task should mimic challenges the candidate will face in their day-to-day work. The complexity and type of work assignment will differ based on functional area and career level.
- The assignment should incorporate similar constraints and resources the candidate would encounter on the job.
- Think of this like work you are doing together, just as you would with a team member. The candidate’s job is to present you with their ideas and approach. Your job is to give feedback, ask questions, and evaluate their thinking. Choose a work sample that is engaging and interesting for both of you. (Remember, the candidate is also evaluating what it’s like to work with you.)
For the research behind why work sample testing is so effective (and real examples from association and nonprofit executive searches in varying functional areas and career levels), see “SA Perspectives: Work Sample Testing.”
Before you give a work sample to candidates, test it internally. Ask:
- Does the complexity of the task reflect the demands of the role and the level of seniority? Does the work sample closely resemble scenarios the candidate will encounter on the job?
- Is it reasonable that a candidate could complete it within the given timeframe? (We recommend no longer than two hours of work with a week to complete.)
- Will the work sample provide insights that interview questions might not?
- Have you defined your assessment criteria? Are there multiple viable approaches to the work? What follow-up questions will you ask each candidate to evaluate them consistently and fairly?
- Run it through ChatGPT. What’s lacking in the AI’s answer that you’d like to hear from candidates? Can you add any context that requires a deeper response than what the AI delivers?
Start a Discussion
It can be easy to view the work sample as an end in itself, asking candidates to turn something in and stopping there. But a deliverable tells you nothing about how a candidate arrived at their conclusion, or how they will pivot when something new comes up. ChatGPT can create a SWOT analysis and emerging tools like Microsoft CoPilot can create presentations, but neither can engage in a discussion about a work challenge. You’ll get far more insight into how a candidate solves problems and whether they understand the position from a discussion about the work.
Ask candidates to, “Come prepared to discuss how you would handle (some challenge).” During the discussion, ask them to walk you through their approach. Follow up with detailed questions and challenge their ideas. You’re not looking for the right answer; you’re looking for thought processes that will benefit your team.
Suggested prompts include:
- Walk me through your thought process. What were your key objectives? How did you prioritize your steps?
- Which aspects of your past experience were most relevant in helping you complete this work?
- If you had more time or resources, what additional steps would you have taken?
- How would you present your approach to someone who is unfamiliar with this work?
- How would you get buy-in from different stakeholders or team members?
- I (x variable) changed in the scenario, how would you shift your approach, if at all?
Reflect on the Discussion
After your discussion with the candidate, ask yourself or your team:
- Was the candidate prepared?
- Did they show an understanding of the context of our organization (constraints or opportunities unique to us?)
- Did they make reasonable assumptions (even if not always accurate)?
- Did they ask about and include the perspectives of key stakeholders?
- Would they work well with the executive team (if applicable)? Would their work style and perspective add value?
- Would they effectively engage with managers and staff? How did they respond to inquiry and feedback?
- What support would they need to be successful in this role?
Treat the Work Sample Like Actual Work
Imagine your supervisor asks you to discuss a critical issue at an upcoming meeting. You’d likely research, pull from several sources and your own experience, and develop some ideas. It doesn’t really matter what resources you used to prepare; what’s more important is how you present to the team, wrestle with holes in your thinking, and tweak your approach. That’s how we get work done. And that’s what a great work sample test looks like. The effectiveness of this step in your hiring process hinges on your involvement with the candidate; the discussion is the secret sauce. Focus on that, and your process will always be ChatGPT-proof.
(If you’d like to learn how Results-Based Hiring® practices like work sample testing can help you find and retain exceptional candidates who will drive impact in your nonprofit or association, contact the Staffing Advisors team. We would love to work with you.)
- For a step-by-step guide to interviewing executives using a skills-based approach, read our Employer Guide to Interviewing.
- To see how work sample testing fits within your hiring process, read The Principles of Competency Driven Interviewing.
- For a peek at how behavioral economist Steve Levitt used work samples to hire for his team at the University of Chicago, read Work Sample Testing in the Wild.