Practice for pymetrics-style games

Some firms use short browser games to build a trait profile rather than a score. CaseTutor's six original games cover the same kinds of tasks candidates report, and turn your play into a seven-trait profile compared with a consulting benchmark. It takes about twenty minutes.

Start free practice

A free CaseTutor account is required. Your first full attempt is free; further attempts are part of a membership.

The format

Six-game battery

Time:
22 minutes
Format:
6 short games, played in order
Tools:
Keyboard or mouse

Pump and bank balloons, recall growing digit strings, respond or hold back at a fixed pace, tap in timed bursts, solve disc puzzles in few moves, and draw from card decks that change partway through.

What test day looks like

There is no content to revise here, which is exactly why the page is worth reading. What you can control is knowing each game's rules before the clock starts, and playing in conditions where your reactions are actually yours.

Format
Short browser games, about two to three minutes each
Reported battery
Around 12 games, 25 to 35 minutes in total
What it produces
A trait profile against the firm's benchmark, not a pass mark
Attempts
Generally one. Results are reported as reusable by other employers for about 330 days
Breaks
Reported as one sitting; you cannot pause inside a game
Our coverage
Six original games, about 22 minutes
  • The full battery includes games we do not simulate, such as face reading, arrow direction and length estimation. Our six cover task types including: risk, learning from reward, working memory, impulse control, effort, planning and learning under risk.
  • pymetrics is now part of Harver. Firms that use it calibrate their own benchmark, so the same play can read differently at two employers. Bain and BCG have both used it in some offices; neither has it as the default screen in 2026.

Original worked example

A worked example: why the balloon game is not about the money

An original walkthrough of the trait read, using our own numbers. It shows why practising to beat the game tends to make the profile worse, not better.

Two candidates play ten balloons. Each pump adds $0.05 to the balloon and a balloon can burst at any point after the fourth pump. Banking keeps whatever is in the balloon; a burst pays nothing. Candidate A banks at eight pumps every time and bursts once. Candidate B banks at four, then ten, then four, then ten, alternating to the end, with no bursts.

Who made more money, and does the money tell you who took more risk?

  1. Candidate A

    Nine successful banks of eight pumps is 72 pumps, or $3.60. The burst pays nothing, so the total is $3.60 from ten balloons.

  2. Candidate B

    Five balloons banked at four pumps is 20 pumps, five at ten pumps is 50, so 70 pumps and $3.50. Within 3% of A, with a completely different pattern.

  3. Read the pattern, not the payout

    A uses the same banking point each time, while B alternates between lower and higher choices. The examples show different patterns, but do not establish how an employer would interpret them.

  4. Why gaming it backfires

    Employers do not publish the full scoring approach. Avoid guessing the profile they want; focus on understanding the instructions and responding naturally.

The answer

The money is level, within 3%, and the profiles are not. Play the way you would play on the day. The only thing worth rehearsing is the interface, so that working out the controls does not read as hesitation.

What it is worth under the clock. Reaction and attention games record timing, not just answers. A lag while you find the button is indistinguishable from a slow decision, which is why one cold run before test day is worth more than five attempts at a better score.

How your attempt is graded

  • Your play produces seven trait signals: risk tolerance, learning from rewards, working memory, impulse control, effort, planning and learning under risk.
  • Each trait is compared with a consulting benchmark range. The report is a fit profile, not a grade.
  • Every result is worked out from what you actually did in each game.

Tips for the real thing

  • There is no 'right' profile. Play naturally, the way you would on the day.
  • Use a keyboard if you can. The tapping game is hard to play fairly on a touch screen.
  • Play somewhere quiet. Several games measure reaction and attention.

A seven-day prep plan

This plan is about removing noise, not improving a score. Every day either teaches you a rule or removes a reason for your play to be unrepresentative.

  1. Day 1

    Play once, cold

    The full battery with no pausing. Note where the interface, rather than the task, slowed you down.

  2. Day 2

    Read the profile, not a score

    Find the two traits furthest outside the benchmark band. Ask honestly whether they describe you; if they do, there is nothing to fix.

  3. Day 3

    Mechanics only

    Replay the two games you found confusing. The goal is to stop thinking about the controls, not to change the outcome.

  4. Day 4

    Set up the room

    Laptop, keyboard, full screen, phone away, quiet. The tapping and reaction games are unfair to play on a touch screen or in a busy room.

  5. Day 5

    Timing rehearsal

    Twenty-two minutes in one sitting, breaks only between games, to get used to sustained attention rather than short bursts.

  6. Day 6

    Rest

    No games. Reaction time and attention are measured, so a tired run is a worse profile, not extra practice.

  7. Day 7

    Play once, properly

    If your invitation names Harver or pymetrics, this is the run that counts. Play naturally and let the profile stand.

Frequently asked questions

Can you practise for pymetrics games?

You can't learn right answers, because the games measure traits rather than knowledge. Practice helps you know what each game asks, so nerves and confusion don't distort your profile.

Are these the real pymetrics games?

No. All six games are original, with their own art and mechanics. They cover similar kinds of tasks, and the benchmark is CaseTutor's own.

How long does it take?

About 22 minutes for all six games.

Is the practice test free?

Your first full attempt at each module is free with a CaseTutor account, including the full report. Further attempts are part of a CaseTutor membership.

Sources and how current this is

The game count, timings and reuse window are candidate-reported and agree across the sources checked on 17 November 2026. Harver does not publish a game list or scoring. Which employers use the battery, and with what benchmark, changes by office and year.

Page reviewed 17 November 2026. Format details change between cycles; your recruiter's instructions are the authority.

Your first attempt is free with an account

Create a free CaseTutor account, then sit a full, timed simulation and get a scored report on where you lost points.

Create a free account

Preparing for the whole process?

Bain interview prep →

Other online assessments

CaseTutor is an independent, unofficial resource and is not affiliated with, endorsed by, or sponsored by McKinsey & Company, BCG, Bain & Company, Harver/pymetrics, SHL, Sova Assessment, TestGorilla, or HireVue. All trademarks are the property of their respective owners. This is original practice software that emulates the format and skills of third-party assessments; it does not reproduce any assessment’s questions, artwork, or scoring.