HireVue Games Explained: Every Game and How to Score Well
HireVue Games are a 16-task adaptive cognitive battery used by JPMorgan, Goldman Sachs, Bain, BCG, Capital One, IBM, Microsoft, and Amazon. Each game runs about 3 minutes with 15 to 20 seconds per level, and the AI rewar
The honest answer is that HireVue Games are not "fun mini-games" and they are not a personality fluff layer either. They are a 16-task adaptive cognitive battery, each running roughly 3 minutes, with 15 to 20 seconds allowed per level inside that. Companies like JPMorgan, Goldman Sachs, Citi, Bain, BCG, Capital One, IBM, Microsoft, and Amazon use the games to screen candidates before any human looks at a CV. If you treat them like Candy Crush you will score in the bottom third. If you treat them like a 3-minute timed cognitive test, which is what they are, you have a fair shot.
Quick takeaways
- 16 HireVue Games exist in the catalog. Most assessments use 4 to 8 of them, picked by the employer.
- Each game runs about 3 minutes, with 15 to 20 seconds per level. Adaptive difficulty: do well early and harder, higher-value items unlock.
- Speed and accuracy both count. A wrong fast answer scores worse than a right slow one.
- HireVue phased out facial analysis in 2021 after public criticism. The AI now scores verbal content and behavioral patterns, not your face.
- You will not see your score. Only the employer and recruiter see your percentile.
- The games feed a percentile against the role's reference group, not a raw number.
- Cutoffs are set by the company. For investment banking and consulting screens, candidates report needing roughly the 70th percentile or higher to advance.
What HireVue Games actually are
HireVue Games are a set of short, timed cognitive and behavioral tasks delivered inside HireVue's assessment platform. They are not the video interview. They sit in front of, or alongside, the on-demand interview as a pre-screen filter. A typical HireVue assessment funnel looks like this: candidate gets an invitation link, completes 4 to 8 games in roughly 25 to 35 minutes, then either gets cut, gets moved into the on-demand video interview round, or both happen in the same session.
The catalog has around 16 games, but no employer uses all of them. Capital One's analyst screen leans on Numerosity, Flashback, and PortraitXT. JPMorgan and Goldman Sachs investment banking screens lean on Numerical Reasoning, Verbal Reasoning, and one or two cognitive memory games. Bain and BCG use the games for early consulting hires, often paired with Pymetrics-style behavioral data. Microsoft and Amazon use the games for early-career engineering roles where cognitive aptitude needs to be measured fast.
The framing matters because candidates often arrive expecting Pymetrics. Pymetrics is a separate gamified product, sold by Harver, and uses a different scoring philosophy (neuroscience-style trait matching against high-performer profiles). HireVue Games are cognitive performance tests dressed up in a game UI. The visual language is similar; the scoring philosophy is not. If you have prepped for Pymetrics, do not assume the same approach works here.

Every game in the catalog, what it actually measures
Below is the working catalog. Different sources count between 12 and 18 games depending on whether they include the verbal, numerical, and logical reasoning sub-tests (which HireVue lists under the same "games" umbrella in 2026). The list below covers what you are most likely to see.
Numerosity. Visual processing under time pressure. You see a 4x4 grid of squares, each containing dots, and you pick the square that holds the target number of dots before the screen changes. Roughly 3 minutes. The score rewards correct hits, with speed as the tiebreaker. This is the single most common game in finance and analytics screens.
Flashback. Working memory. You see a shape, then it disappears, then a new shape appears. You decide whether the new shape is the same as the previous one. Then a new shape, then you compare against the one before that. The "lag" is what makes it hard: at the higher levels you are comparing against a shape from two or three steps back, not the last one.
Shapedance. Working memory plus pattern detection. Several cubes with shapes inside, arranged in a pattern. You identify the pattern logic and complete it. 3 minutes total.
Digitspan. Short-term memory span. You see a sequence of digits and letters, the screen clears, and you type the sequence back. Sequences start at 4 to 5 items and grow until you fail. Your "span" is the longest you hold accurately.
Singularity. Visual processing speed and attention. Items flash on screen and you decide whether they match a target rule. 3 minutes. Pure speed game.
PortraitXT. Personality and workplace preferences. About 50 statements, you rate agreement on a Likert-style scale. Roughly 5 minutes. Not cognitive; this feeds the role-fit component.
Picturequiz. Recognition and recall. You see images and answer questions about what was shown. Tests short-term visual memory.
Verbal Reasoning. Argument analysis. You read short passages and answer whether statements logically follow. Up to 12 minutes depending on the employer's configuration. This is closest to a Watson-Glaser style question.
Numerical Reasoning. Data table interpretation. You see a table or chart, then answer percentage/ratio/trend questions about it. Up to 12 minutes. The investment banking screens lean hard on this one.
Logical Reasoning. Abstract pattern completion. Shapes follow a rule, you pick the next one in the sequence. Up to 12 minutes.
The remaining 6 to 8 games in the broader catalog (Quickfire, Mood Match, Pathway, etc.) appear in specific employer configurations and are variants of the same cognitive primitives: memory span, pattern recognition, reaction time, working memory.
How the AI actually scores you
This is where most candidate myths live, so the answer needs to be specific. HireVue's AI in 2026 does three things with your data: it scores your cognitive game performance using item response theory (correct hits weighted by question difficulty, not raw counts), it scores your behavioral patterns (response time variance, level progression curves, where you slow down), and for the video interview portion it scores the verbal content of your answers using natural language processing.
What HireVue's AI no longer does: facial expression scoring. HireVue retired the visual facial analysis component in early 2021 after research published by AI Now Institute and reporting from The Washington Post raised validity concerns. The 2026 product scores spoken content and language patterns, not your micro-expressions, not your smile rate, not your eye contact.
Inside the games, the scoring logic looks roughly like this: each game produces a raw score, the raw score is converted to a percentile against the role's reference group (other candidates who have taken the same configuration for similar roles), and the percentiles are bundled into a composite that the recruiter sees. The recruiter does not see your raw answers. They see a percentile rank, often with a recommended cutoff drawn at the 50th or 70th percentile depending on the role.
Speed matters but does not dominate. A wrong fast answer hurts more than a right slow answer, because the underlying item response model penalises low-confidence mistakes more than slow caution. The candidate myth that "rush through to finish more items" is wrong. The right approach is to be deliberate enough to be accurate, fast enough to clear most of the items.
The chart below shows how this plays out across the game catalog, what each one measures, how long you get, and what the AI specifically rewards.

Score targets by employer tier
Most candidates do not know what they are aiming for, so they aim for "do my best" and find out at the rejection email. The defensible targets are roughly:
| Employer tier | Example employers | Cutoff percentile (reported) | What this means |
|---|---|---|---|
| Bulge bracket banking | JPMorgan, Goldman Sachs, Citi | 75th to 90th | Above average on cognitive games, strong on numerical reasoning |
| Tier 1 consulting | Bain, BCG, occasionally McKinsey | 70th to 85th | Strong across cognitive, especially logical reasoning |
| Big tech early career | Microsoft, Amazon | 60th to 80th | Above-average cognitive, balanced across games |
| Financial services (analyst) | Capital One, Wells Fargo | 60th to 75th | Solid on Numerosity and numerical reasoning |
| Fortune 500 generalist | Various | 50th to 65th | Pass the cognitive floor, role fit matters more |
| High-volume retail/ops | Various | 30th to 50th | Cognitive games used as soft filter, behavioral games matter more |
These are reported figures based on candidate-side write-ups, recruiter conversations, and what HireVue publishes about role-based benchmarks. HireVue does not publish exact cutoffs and companies treat them as confidential. Use the table as orientation, not as a guarantee.
How to score well on each game
The game-by-game tactics that actually move scores:
Numerosity. Train your subitizing range. Subitizing is the brain's ability to instantly count up to 4 or 5 dots without counting one by one. Most people max out at 4. With practice you can reliably hit 6 or 7. Free apps and the official HireVue practice flow both help. In the test itself, scan the four squares in a Z pattern, not left-to-right; the Z pattern reduces revisits.
Flashback. This is the game where preparation gap shows the most. Practice the 1-back, 2-back, 3-back working memory tasks (often called the "n-back" in cognitive science). Free apps exist (Brain Workshop, Dual N-Back). 15 minutes per day for a week measurably improves your span.
Shapedance. Pattern recognition. Less trainable in a week, more trainable across months. Strategy in the test: do not get stuck. If the pattern logic does not resolve in 8 to 10 seconds, guess and move on. The adaptive engine penalises stalled items.
Digitspan. Use chunking. Group digits into 3s or 4s, the way you remember phone numbers. Average untrained span is 7 plus or minus 2. With chunking you can stretch to 10 or 11.
Singularity. Pure reaction time and discrimination. Calibrate by doing the practice round at full focus. Caffeine 30 to 45 minutes before, no later. Hydration matters more than candidates think.
PortraitXT. Be consistent. The 50-statement structure includes deliberate near-duplicates to catch inconsistent answering. If you describe yourself as "detail-oriented" in statement 7 and "big-picture, not into details" in statement 41, the AI flags you as low-reliability. Pick a coherent self-description and hold to it.
Verbal Reasoning and Numerical Reasoning. These two are scored like Watson-Glaser and SHL respectively. The honest answer for these is timed practice. Two to three weeks of 30-minute daily timed practice moves most candidates from the 40th to the 70th percentile.
Logical Reasoning. Patterns repeat. Categories include rotation, reflection, addition/subtraction of elements, color change, and shape morphing. Drill the category catalog and the test becomes pattern-match, not novel reasoning.
A 5-day prep plan that fits the games
Five days is realistic for most candidates. The plan below assumes 60 to 90 minutes per day.
Day 1 (60 min). Take a full untimed sample test on a reputable practice platform. Note which games drained you and which felt natural. Read the official HireVue candidate guide so you know which games this specific employer uses (the invitation email usually names them, or the practice link configures the same set).
Day 2 (90 min). Drill working memory. 20 minutes of n-back app practice. 20 minutes of digit-span chunking practice. 30 minutes of Numerosity-style dot-counting drills. 20 minutes reviewing how scoring works so you stop trying to game it.
Day 3 (90 min). Drill reasoning. 45 minutes of numerical reasoning timed practice (data tables, percentage change, ratios). 45 minutes of verbal reasoning practice (inference vs assumption, the same logic the Watson-Glaser tests).
Day 4 (60 min). Full timed simulation. Use the configuration the employer specified. Treat it like the real thing, including environment: quiet room, full charge, no notifications. Score yourself honestly.
Day 5 (60 min). Light review of your two weakest games from Day 4. No new material. Sleep early. Test day is for execution, not learning.
Common mistakes that tank scores
The recurring patterns from candidates who score poorly:
Treating speed as the only variable. The adaptive engine rewards accuracy weighted by difficulty. Rushing past medium-difficulty items to get to "more questions" leaves easy points on the table. Aim for 85% to 90% accuracy first, speed second.
Skipping the practice round. Every HireVue game has a short practice round before the timed round. Candidates skip it to save time. The practice round calibrates your input lag and your tap accuracy. Skipping it wastes 5 to 10 score points across the assessment.
Wrong device. HireVue Games work on desktop, tablet, and phone, but the games are calibrated for desktop input. Tap latency on phone differs from mouse latency on desktop. Take the assessment on the device the employer's reference group used, which is almost always desktop.
Background distractions. The behavioral signal includes response time variance. If you pause to look at your phone for 12 seconds in the middle of Flashback, your variance signature flags as low-engagement. Even if the answers are correct, the composite score drops.
Over-rehearsed video interview answers. This is for the interview portion paired with the games. The AI scores verbal content for relevance and structure (STAR-style). Memorized scripts get flagged as low-naturalness. Practice the structure, not the words.
FAQ
Can you practice HireVue Games for free?
Yes, partially. HireVue itself offers a candidate practice flow inside the assessment platform that mirrors a subset of the games, mainly the cognitive ones (Numerosity, Flashback, Digitspan). Beyond that, third-party prep providers offer paid practice. The n-back, digit-span, and reasoning components can be practiced for free using cognitive science apps (Brain Workshop for n-back, free SHL and Watson-Glaser sample tests for reasoning).
How long does the full HireVue Games assessment take?
Most employer configurations land in the 25 to 35 minute range total, covering 4 to 8 games. Some banking screens stretch to 45 minutes when they add the on-demand video interview in the same session. The invitation email tells you the expected duration, and HireVue does not penalize you for going slightly over inside the per-game limits, only for being slow on each individual item.
Do you see your HireVue Games score?
No. HireVue does not show you your score at the end of your assessment. Your percentile rank and game-by-game breakdown go to the employer and recruiter. You will see a generic "thank you, we will be in touch" screen. Some employers share aggregate feedback after a rejection, most do not.
Does HireVue use AI to scan my face?
Not as of 2026. HireVue retired the facial analysis component of its AI scoring in early 2021 following research and media scrutiny on validity. The current AI scores verbal content of spoken answers (in the interview portion), response patterns in the games (timing, accuracy curves, adaptive difficulty navigation), and behavioral consistency across the assessment. It does not score your face, your smile, or your eye contact.
Can you retake HireVue Games?
Only if the employer invites you to. HireVue itself does not allow self-initiated retakes. If you bombed an assessment and want another shot, you can ask the recruiter directly. Some employers allow one retake within 6 to 12 months for the same role. Most do not.
Which is harder, HireVue Games or Pymetrics?
Different products, different difficulty profiles. Pymetrics is harder to "study for" because it scores trait fit against a high-performer profile rather than raw cognitive performance. HireVue Games are harder for candidates with low working memory baselines because the cognitive games (Numerosity, Flashback, Digitspan) reward measurable cognitive ability that you cannot fake. Most candidates find HireVue Games harder if they have not prepped, and Pymetrics harder to gain on with preparation.
Do HireVue Games count more than the video interview?
It depends on the employer. For bulge bracket banking and Capital One analyst screens, the games dominate; the interview is a sanity check. For consulting (Bain, BCG), the interview content carries more weight, but a poor games score still filters you out before the interview is reviewed. For Fortune 500 generalist roles, the interview content carries the day, and the games are a soft filter.
What if I have a disability that affects my performance?
HireVue offers accommodations through the employer's recruiting process. Common accommodations include extended time per item, alternative formats for the cognitive games, and bypass options for specific games. The accommodation request goes through the employer, not HireVue directly. Request it before the assessment link expires.
Related on PrepClubs
- Pillar. The HireVue Hub: assessment, interview, and games. Everything HireVue, in one place.
- Deep practice. Practice HireVue Games with timed simulations. $39 one time. Pass Guarantee.
- Format. HireVue in 2026: assessment, interview, games, and companies. The brand-level overview.
- Interview side. HireVue Interview: format, questions, and how to pass. For the on-demand video portion.
- AI scoring. HireVue AI: how the algorithm actually scores your interview. The 2026 algorithm explained.
Practice on PrepClubs
Practice HireVue Games at the pace HireVue actually uses.
The free HireVue practice flow shows you the format. It does not give you the timing pressure, the adaptive difficulty curve, or the variety of game configurations that real employer assessments use. PrepClubs runs timed simulations across all 10 main games at the 15-to-20-second-per-level cadence employers actually deploy, with score reports that match the percentile bands above. $39 one time. Pass Guarantee.
FAQ


