Game-Based Assessments 2026: Arctic Shores, Pymetrics & More
What game-based assessments actually measure, how Arctic Shores and Pymetrics work, which employers use them, what the science says — and what you can genuinely do to prepare.
What Are Game-Based Assessments?
Game-based assessments are psychometric tools disguised as mini-games. Instead of answering multiple-choice questions about shapes or charts, you play a series of short interactive tasks — catching falling items, tapping balloons, sorting objects under time pressure — while the platform records hundreds of behavioural data points: reaction time, decision patterns, risk tolerance, attention shifts, and error recovery behaviour.
These behavioural traces are processed by machine learning models trained on data from high-performing employees at the hiring organisation. The output is a profile of cognitive and personality traits — not a single score, but a multi-dimensional match against what the employer's data says predicts success in that specific role.
The key conceptual shift: in an SHL test, your score depends on whether your answers are correct. In a game-based assessment, your profile depends on how you play — the speed of your decisions, how you respond to uncertainty, whether you take risks when reward is unclear, how you recover from mistakes. There are no universally "correct" answers. The platform is measuring your cognitive style, not your knowledge.
Arctic Shores: Platform & Games
Arctic Shores is a UK-based game-based assessment platform founded in 2013 and used by employers including PwC (some streams), AstraZeneca, Deloitte (select programmes), Unilever, and various financial services firms. The Arctic Shores assessment typically involves 10–18 mini-games completed over 25–35 minutes, producing a trait profile across cognitive and behavioural dimensions.
Arctic Shores Mini-Games and What They Measure
🎯 Target Practice
Tap moving targets of varying sizes and speeds. Measures: processing speed, accuracy under pressure, attention to detail, motor precision.
🔵 Balloon Pump
Pump a balloon to increase points — but it bursts if pumped too far. Measures: risk tolerance, reward sensitivity, impulse control, decision-making under uncertainty.
📦 Object Sorting
Sort falling objects into categories that change mid-task. Measures: cognitive flexibility, ability to shift rules quickly, error recovery, working memory.
🃏 Card Games
Make decisions based on partial information revealed incrementally. Measures: probabilistic reasoning, updating beliefs under uncertainty, patience vs. impulsivity.
🔢 Number Memory
Memorise and recall sequences of increasing length. Measures: working memory capacity, concentration, performance under cognitive load.
⏱️ Timing Tasks
Estimate time intervals or tap at precise moments. Measures: temporal awareness, precision, ability to regulate behaviour to a rhythm or target.
🔄 Rule Learning
Discover hidden rules by trial and error within a fixed number of attempts. Measures: learning agility, pattern recognition, hypothesis testing, efficiency.
😊 Emotion Recognition
Identify emotions expressed in brief facial images. Measures: social perception, empathy, emotional intelligence, interpersonal awareness.
Arctic Shores maps game behaviour to a trait framework typically covering: Cognitive Processing (speed, accuracy, working memory), Learning Agility (rule acquisition, flexibility), Emotional Resilience (error recovery, frustration tolerance), and Social Orientation (empathy, collaboration markers). The specific trait model varies by employer configuration.
Pymetrics: Platform & Games
Pymetrics (now part of Harver) is a US-based game-based assessment platform that uses 12 neuroscience-based games completed over approximately 25 minutes. Pymetrics was adopted by a number of major employers — including Unilever, LinkedIn, Kraft Heinz, and some financial services firms — for graduate and entry-level hiring, often as a replacement for traditional personality questionnaires.
EY's strengthened assessment process no longer uses Pymetrics as a mandatory component for all streams as of recent cycles. If you see Pymetrics mentioned in EY preparation guides, verify against the current year's EY careers website before preparing specifically for it. The Pymetrics games below are still directly relevant for all other current Pymetrics users.
The 12 Pymetrics Games
🎈 Balloon Game (BART)
Pump a balloon for increasing reward; it may burst. Measures: risk appetite, reward sensitivity, impulsivity. Directly analogous to Arctic Shores Balloon Pump.
💰 Money Exchange (Ultimatum Game)
Decide whether to accept or reject offers of split sums. Measures: fairness sensitivity, social preference, pro-social behaviour, negotiation style.
🃏 Card Sorting (Iowa Gambling Task)
Choose cards from decks with varying reward/punishment profiles to maximise return. Measures: learning under uncertainty, decision-making quality, risk learning.
⬆️ Arrows
Respond to arrow direction while ignoring distracting arrows. Measures: attention control, inhibitory control, processing speed, focus.
🔢 Digits Forward/Backward
Recall number sequences in forward and reverse order. Measures: working memory, concentration, cognitive capacity.
🟡 Lengths
Judge which of two lines is longer under time pressure. Measures: processing speed, perceptual precision, decision speed.
😄 Faces
Identify the emotion expressed in rapidly changing facial images. Measures: emotional intelligence, social perception, empathy markers.
⏱️ Hard or Soft (Stop Signal)
Respond to stimuli but inhibit response when a stop signal appears. Measures: impulse control, inhibitory control, ability to stop initiated actions.
💸 Money Exchange 2 (Trust Game)
Decide how much to send to a partner who may or may not reciprocate. Measures: trust propensity, social risk-taking, cooperative behaviour.
🔵 Towers of London
Move coloured discs to achieve a target arrangement in minimum moves. Measures: planning ahead, spatial reasoning, strategic thinking, problem-solving.
🎯 Keypress
Tap as fast as possible during target intervals; withhold during non-target intervals. Measures: motor speed, vigilance, sustained attention.
🌊 Penguin Escape
Guide a character through obstacles requiring quick directional decisions. Measures: spatial reasoning, planning, reaction time, cognitive flexibility.
Other Game-Based Platforms
| Platform | Parent / Status | Format | Notable Users |
|---|---|---|---|
| HireVue Game-Based | HireVue (acquired Shaker International) | 6–8 mini-games embedded in or alongside video interview; 15–20 min | Goldman Sachs (some streams), JPMorgan (some streams), large US banks and insurers |
| Unilever Future Leaders League | Unilever proprietary (built with Pymetrics + Arctic Shores elements) | Custom game suite; 20–25 min; feeds directly into UFLP online screen | Unilever (UFLP only) — used alongside Korn Ferry Talent Q aptitude tests |
| Revelian (Criteria) | Criteria Corp (Australia/NZ) | Game-based cognitive + personality; 25 min | Large Australian employers including banks and utilities; increasingly common in ANZ |
| Cognify (Criteria) | Criteria Corp | 3 game-based cognitive tasks; 15 min | Mid-tier employers globally — faster/cheaper alternative to full game suites |
The Science: What Game-Based Assessments Actually Measure
The psychometric constructs measured by game-based assessments map onto three well-established frameworks from cognitive and personality psychology. Understanding these frameworks helps you understand both what the assessment is doing and why preparation in the traditional sense is limited.
| Construct | What It Means | Games That Measure It | Predictive Validity |
|---|---|---|---|
| Cognitive Processing Speed | How quickly and accurately you process and respond to information | Arrows, Keypress, Lengths, Target Practice | Strong predictor of job performance across roles — one of the most robust findings in I/O psychology |
| Working Memory | How much information you can hold and manipulate simultaneously | Digits, Number Memory, Towers of London | Strong predictor of learning speed, complex task performance, and trainability |
| Inhibitory Control | Ability to suppress impulsive responses and withhold actions | Stop Signal, Arrows (with distractors), Sorting | Predicts self-regulation, error rate, and conscientiousness markers |
| Risk Tolerance | How you balance potential reward against potential loss under uncertainty | Balloon Game (BART), Card Sorting (Iowa) | Predicts behaviour in ambiguous or high-stakes decisions; calibrated to role type |
| Learning Agility | How quickly you adapt when rules change or new patterns emerge | Card Sorting, Rule Learning, Object Sorting | Predicts performance in new or changing environments; valued by consulting and fast-growth employers |
| Social / Emotional Intelligence | How accurately you read and respond to social and emotional cues | Faces, Money Exchange, Trust Game | Predicts performance in roles requiring stakeholder management, client-facing work, or team leadership |
A high risk tolerance score is not universally better than a low one. Employers calibrate the algorithm against their existing high performers. A consulting firm may weight learning agility and cognitive speed most heavily. A risk management role may penalise high risk tolerance. You cannot game the scoring by aiming for an extreme on any trait — the "right" profile is role-specific and unknown to candidates.
Which Employers Use Game-Based Assessments?
| Employer | Platform | Stage in Process | Role in Selection |
|---|---|---|---|
| PwC | Arctic Shores | After online aptitude tests; before first interview | Supplementary — combined with SHL scores to form overall screen profile |
| AstraZeneca | Arctic Shores | After application; before video interview | Primary online screen for some graduate programmes |
| Unilever UFLP | Pymetrics + custom game suite | After Talent Q aptitude tests | Mandatory — replaces traditional personality questionnaire; feeds AI hiring recommendation |
| Kraft Heinz | Pymetrics | Early stage online screen | Primary online screen; well-documented Pymetrics user |
| Pymetrics | After application | Early stage screen for some roles | |
| Goldman Sachs | HireVue Game-Based (select streams) | Embedded in or alongside HireVue video interview | Supplementary to aptitude + video; contribution to overall score varies by stream |
| Deloitte | Arctic Shores (select programmes) | Supplementary to Immersive Online Assessment | Trait data feeds into overall profile; not primary filter |
| Various AU employers | Revelian (Criteria) | After application or alongside aptitude tests | Cognitive + personality game suite; increasingly adopted in Australian market |
Common Myths Debunked
"Just play as fast as possible — speed is everything."
Speed-accuracy tradeoffs are tracked explicitly. Frantically clicking without accuracy produces a different (and often worse) profile than measured, accurate responses. The platform measures both dimensions independently.
"Aim to be risk-averse — that's what employers want."
The "right" risk profile is role-specific. Sales and S&T roles may prefer higher risk tolerance; risk and compliance roles may favour conservative profiles. There is no universally optimal risk score across all roles.
"These games are easy — you don't need to prepare."
The games are cognitively simple but psychometrically sophisticated. Candidates who are unfamiliar with game formats, playing on an unfamiliar device, or in a distracting environment systematically underperform their natural profile due to situational noise — not low cognitive ability.
"You can cheat by deliberately performing badly on the balloon game to appear risk-averse."
Deliberate manipulation is detectable. The platform collects hundreds of behavioural signals per game — not just the final outcome. Patterns inconsistent with natural play (e.g. artificially stopping at exactly the same pump count every time) are flagged as atypical. Platforms have trained on millions of authentic play sessions.
"Game-based assessments have no predictive validity — they're just a gimmick."
Independent research has found predictive validity for cognitive game measures comparable to traditional psychometric tests. The BART balloon task and Iowa Gambling Task have decades of research behind them. The main legitimate critique is construct validity for the personality/social trait measurements, which are newer and less peer-reviewed.
What You Can Actually Do to Prepare
Game-based assessments are specifically designed to be resistant to conventional preparation — you cannot revise the "right answers" in the way you can for SHL Numerical. However, there are meaningful things you can do to ensure your natural cognitive profile is accurately captured rather than distorted by situational factors.
- Use a desktop or laptop, not a mobile device. Game-based assessments are optimised for mouse/trackpad interaction. Reaction time, tap accuracy, and navigation precision are all measured — and all perform worse on a touchscreen. Using a larger screen also reduces cognitive load from the interface, letting your actual ability come through.
- Ensure a distraction-free environment. The tasks measure sustained attention and cognitive performance. Background noise, interruptions, or multitasking between games creates variability in your data that the platform interprets as trait noise — it may lower the confidence of your trait estimates or produce a profile that doesn't reflect your natural performance.
- Play each game fully engaged, at your natural pace. Don't try to game the scoring. The most authentic representation of your cognitive style is the most stable signal. Artificial behaviour (pumping the balloon exactly the same number of times every round, clicking at an unnaturally uniform pace) is detectable and interpretable as low authenticity.
- Familiarise yourself with the game interfaces beforehand. Most platforms offer practice rounds or publicly available demo versions. Spending 10–15 minutes with the interface before your real assessment removes the cognitive load of "what do I click?" from your actual performance. This is meaningful, legitimate preparation.
- Get adequate sleep and sit the assessment when cognitively fresh. Processing speed and working memory — two of the most heavily measured dimensions — are acutely sensitive to sleep deprivation. Sitting a game-based assessment at 11pm after a long day will produce measurably different results from sitting it at 10am after adequate sleep. This is not a minor effect.
- Focus your preparation energy on the aptitude tests that bookend it. For almost every employer using game-based assessments, the SHL or Talent Q aptitude tests are the primary quantitative filter. The game assessment is supplementary, trait-based, and harder to influence. Your preparation ROI is significantly higher on the aptitude tests — especially when applying to PwC, Deloitte, or Unilever where game assessments supplement rather than replace SHL/Talent Q.
Frequently Asked Questions
Focus Your Preparation Where It Counts Most
Game-based assessments supplement aptitude tests — they don't replace them. Start with the SHL practice that determines whether employers see your application at all.