Recruitment
Soft Skills Assessment That Candidates Can't Cheat
By WiseWorld
Can candidates cheat your soft skills assessment? Compare validity, completion, and prep risk across personality tests, AI interviews, and job-built scenarios.
This article answers four questions:
- Can candidates prepare for every soft skills test used in hiring?
- Which soft skills assessment methods actually predict a good hire?
- Which tests do candidates actually finish?
- Why is a scenario built from your job description the one they cannot look up?
Soft skills assessment in hiring: four questions worth asking
Checking soft skills before the interview is now a normal part of hiring. Most teams run a personality quiz, an AI interviewer, or a skills test on qualified candidates. The problem is that candidates can prepare for almost all of them. About seven in ten job seekers now use AI tools like ChatGPT on applications and interview prep. When everyone can look up how to pass your test, the test stops telling you who will actually do well on the job.
We looked at the research on soft skills assessment in hiring and pulled out four questions every recruiter should be able to answer before choosing a tool.
Can candidates prepare for every soft skills test used in hiring?
People ask "Can candidates cheat with ChatGPT?" as if it were a yes or no question. It is more of a sliding scale. It comes down to one thing: does the test have a right answer that a candidate can find or guess ahead of time?
Here are the common tests, from easiest to prepare for down to hardest.
- A personality quiz. Candidates can sense which answers look good and pick those. This has been true for decades, and these quizzes are also the weakest at predicting who will actually do well.
- A set of fixed questions. Skills tests and scenario quizzes reuse the same questions for every company. Once questions are shared online, candidates can practise them.
- An AI interviewer. Candidates rehearse and re-record until it sounds smooth. It also puts them off. When one company required an AI interview, 38% of candidates dropped out rather than do it.
- A rehearsed interview story. The classic prepared answer. Coaches, online forums, and now AI all hand out the same story templates.
- A live, built-for-you scenario. A conversation that changes based on what the candidate just said, scored against what matters for your job. There is no shared right answer to look up. Candidates can still do well or badly. They just cannot game it, because they never see what you are weighting.
How easy each method is to prepare for or game
A higher bar means more of the answer can be looked up, practised, or handed to AI. A live roleplay built from your own job is the hardest to game.
The chart shows the pattern clearly. Fixed forms sit in the easy-to-game corner. A live conversation built from your own job sits far away from it.
Which soft skills assessment methods actually predict a good hire?
A candidate can prepare for a test and still be wrong for the job. Or struggle on a test and thrive once hired. The question here is different: does the method you use actually predict who will do well in the role?
The clearest answer comes from a 2022 review by Sackett and colleagues in the Journal of Applied Psychology. They pulled together decades of hiring studies and asked a simple question: when someone scores well on a test, do they actually perform well on the job months later?
What they found: the ranking stays much the same as older research, but the numbers are more conservative. Structured interviews and job-like tasks still sit at the top. Personality quizzes sit near the bottom. The gap between them is real.
Why it matters for hiring: most teams choose a soft skills tool because it is fast, familiar, or already in the budget. This research lets you compare those tools on evidence, not habit. If your assessment sits at the weak end of the chart, a polished shortlist may not mean much once the person starts.
How well each method predicts real job performance
Based on decades of hiring research. A taller bar means the method more reliably predicts who will actually do well on the job.
Most teams do not pick tools from this research list. They pick what fits the workflow. The table maps what teams actually run at this stage of hiring against what the research says about prediction.
| Tool | What teams like about it | Where candidates can prepare their way around it |
|---|---|---|
| AI interviewer (e.g. HireVue) | Scales well; used by large enterprises | Answers can be scripted; many candidates drop out; scoring feels like a black box |
| Skills test libraries (e.g. TestGorilla) | Broad; fast to set up | The same questions go to every company, so they can be looked up and practised |
| Personality quizzes (e.g. Predictive Index, SHL) | Familiar to buyers; quick to complete | Weak at predicting job performance; easy to answer the way you think they want |
| Job simulations (e.g. ThriveMap, Vervoe) | Feel like real work; candidates like them | Scenarios come from a shared library, and candidates never see what you weight most |
| Hiring games (e.g. Harver, Pymetrics) | Good for high-volume hiring; engaging | Abstract puzzles, not your job; concerns about fairness for some candidates |
| Coding tests (e.g. Codility, HackerRank) | Strong signal on technical skill | Now easy to complete with AI; say nothing about teamwork or judgment |
| WiseWorld roleplay | Built from your job description; a live conversation; a private space just for candidates | Works best placed after you have checked the basics, like a resume or skills screen |
Which tests do candidates actually finish?
A test that predicts well but nobody completes sends candidates back to your phone screen list. Completion matters as much as accuracy. Here is how candidates rate each method, and how many actually finish.
What candidates are happy to complete
How much candidates like each method, scored out of 7. Tasks that feel like the real job score highest. AI interviewers score lowest.
Candidates rate a realistic scenario well above a personality quiz, and far above an AI interviewer, according to a 2025 review in the International Journal of Selection and Assessment. When a task looks like the actual job, it feels fair. People feel they had a real chance to show what they can do.
How many candidates finish, by method
Share of candidates who complete each type of assessment. Tasks that feel job-related and clear are finished far more often.
Live scenarios and voice tasks finish at 83%. AI interviewers sit at 68%. Form-based tests drop to 64%, based on the 2026 Candidate Voice Report. Every candidate who quits lands back on a recruiter's desk.
Why is a scenario built from your job description the one they cannot look up?
The usual fix for cheating is surveillance. Lock the browser. Track the candidate's eyes. That works for a coding test with one correct answer. It does nothing for soft skills, where what you care about is how someone talks, decides, and handles a tricky moment.
A better fix is to remove the shared answer key entirely. When the test comes from your job description and the conversation changes based on what the candidate says, there is nothing to look up. Here is how that works in practice.
- Paste in your job description. WiseWorld reads it and maps the role to all 44 soft skills. This takes about 5.5 minutes, depending on how much your hiring manager wants to fine-tune.
- Set what matters most. This is your answer key, but it is not a list of correct options. It is a picture of which behaviours count most for this role. Two customer support jobs with the same title can look different here if one team needs someone calm under pressure and another needs a patient teacher.
- Send the scenarios. WiseWorld turns your role into live roleplays. Every scenario comes from your job, not a shared library. Each reply the candidate gives shapes what happens next.
- Get your shortlist. One link goes to your candidates. They complete it on their own time in about 32.5 minutes. You get a ranked shortlist with a match score, a note on gaps, and suggested questions for the interview. Reviewing each one takes about five minutes.
The candidate knows they are being tested on soft skills in a work situation. What they do not know is which skills you value most for this job.
Take a team issue. One person might work through it with active listening: hearing both sides, reflecting back what they heard, and steering toward common ground. Another might use analytical thinking: naming the root cause, weighing tradeoffs, and proposing a clear next step. Both can be strong answers. Neither is automatically wrong. What separates them is fit with the soft skills profile you set for this role. If empathy and collaboration matter most here, the first candidate ranks higher. If structured problem solving matters more, the second one does.
WiseWorld scores every qualified candidate against your profile and ranks who matches best. There is no single script to memorize, only evidence of how each person actually handles the situation. You review the shortlist on your own time instead of booking a manager to run the same conversation with each person.
How do the main options compare on all four questions?
The four questions in this article were four lenses on the same decision: once candidates start preparing with AI, does your soft skills test still tell you something real?
Below is how the main options land on each lens. These are broad categories, not a rating of every vendor. Use them to see where your current tool sits, and what you would trade if you switched.
| Method | Can candidates prepare? | Predicts a good hire? | Do candidates finish? | Built from your job? |
|---|---|---|---|---|
| Personality quiz | Easy to prepare | Weak | Medium (~64%) | Shared everywhere |
| Skills test library | Easy to prepare | Moderate (skills only) | Medium (~64%) | Shared everywhere |
| AI interviewer | Some prep possible | Moderate | Medium (~68%; 38% quit) | Generic questions |
| Job simulation library | Some prep possible | Moderate | High (~83%) | Shared library |
| Hiring games | Some prep possible | Weak | Medium (~64%) | Not your job |
| Manager interview | Hard to game | Strong | Low (needs a calendar) | Built from your job |
| WiseWorld live roleplay | Hard to game | Moderate to strong | High (~83%) | Built from your job |
Where most tools cluster. Personality quizzes and shared skills tests score poorly on prep risk and prediction. They also use the same answer key for every company. AI interviewers and job simulation libraries do better on completion and feel more like work, but candidates can still rehearse against generic or shared content. Hiring games are engaging, but they rarely reflect your actual role.
The tradeoff teams keep hitting. A manager interview scores well on prediction and uses your own bar for what good looks like. It does not scale to every applicant without blocking calendars. A live roleplay built from your job description is the only category here that scores well on all four lenses without a manager in the room each time.
What to do with this. If your current tool sits in the easy-to-prepare column and weak on prediction, assume polished shortlists may not mean much once the person starts. If it finishes poorly, you are paying for dropouts with extra phone screens. If the answer key is shared, the prep economy already has a head start. The goal is not a perfect score on every lens. It is knowing which weakness you are accepting, and whether that matches the role you are hiring for.
What this means for your hiring
Soft skills assessment before the interview is settled. The question now is whether your test still tells you something real. Three checks before you buy or keep a tool:
- Look for a public answer key. If a tool sends the same questions to every company, or scores against a generic personality chart, assume the answers are already online.
- Build the test from the job. Job descriptions already describe how people should behave. Our study of European software engineer job posts found employers write actions, not adjectives. Your soft skills assessment should come from that same job, tuned by your hiring manager.
- Use it after the basics. WiseWorld fits in after a resume or skills check and before the manager interview. It replaces phone screens and weak quizzes, not your coding test or applicant system.
Every tool on the market will keep improving. The one advantage that does not fade is a test candidates cannot rehearse for, because the answer key lives with you and no one else.
Where these numbers come from
We like to be clear about our sources, so you can check them.
- How well methods predict job performance. Sackett, Zhang, Berry, and Lievens (2022) in the Journal of Applied Psychology. We show the results as simple bars rather than technical scores.
- How much candidates like each method. Zibarras, Castano, and Cuppello (2025) in the International Journal of Selection and Assessment, which asked 281 candidates to rate each method out of 7.
- How many candidates use AI. Greenhouse (2025) Workforce and Hiring Report, which found that 67% of US candidates use AI tools when job searching, including applications and interview prep.
- Why candidates drop out of AI interviews. Greenhouse (2026) Candidate AI Interview Report, where 38% of US candidates withdrew rather than complete an AI interview.
- How many candidates finish each test. The 2026 Candidate Voice Report from Recruiting Tech Reviews, based on 2,587 candidates.
- How easy each method is to game. Our own summary based on all of the above and on how these tools work in practice. This one is a guide to the general pattern, not a precise score.
- How the options compare across all four questions. Our synthesis of the sources above, mapped to the four article questions. Broad categories, not a vendor-by-vendor score.
- How WiseWorld works. Checked against the live WiseWorld recruitment product: paste your job description, set what matters, send scenarios, get a ranked shortlist.
- A note on limits. Tool prices and features change often. How well a method predicts performance can vary by role. For high-stakes hiring in regulated fields, always add your own legal review.
More in Recruitment
Latest on the blog
