Harver vs pymetrics

TL;DR: This is not really a versus. Harver completed its acquisition of pymetrics on 11 August 2022 (Harver), so the pymetrics games are now part of Harver's platform rather than a rival to it. Harver sells them as "12+ interactive, gamified experiences" taking "approximately 25 minutes" (Harver). If your invitation says Harver, pymetrics, or both, you are looking at the same family of assessment. What matters is not the brand on the email but which modules you have been sent, because a few are trainable and most are not.

Why "Harver vs pymetrics" is the wrong question

Candidates search this because their invitation mentions one name, a forum thread mentions the other, and the two appear to be alternatives. They are not.

pymetrics built its reputation on short, neuroscience-derived mini-games — balloon-pumping risk tasks, keypress speed, memory span, attention tasks — used to build a behavioural and cognitive profile matched against successful employees in a role.

Harver is a broader hiring platform, best known for high-volume hiring: situational judgement, personality questionnaires, cognitive tests, job simulations, video interviews and screening workflow.

After the acquisition, those game-based modules sit inside the wider Harver offer. So the practical question is not which vendor you face, but which modules your employer switched on — and employers mix and match.

The modules, and which ones you can actually train

This is the distinction that decides how you spend your preparation time.

Behavioural / trait games — not trainable, and do not try. The risk-and-reward tasks, response-style measures and trait games read who you are, not what you can do. There is no high score. Worse, trying to perform a profile tends to produce inconsistency that reads badly against the rest of your results, and profiles are matched to roles rather than ranked, so a manufactured answer can push you toward a role fit you did not want. Play these naturally and consistently. The Arctic Shores balloon game guide explains the same logic for a comparable task in a different battery.

Cognitive games — genuinely trainable. In the pymetrics battery these are Digits (memory span), Arrows (follow the centre or the side arrows depending on colour), Stop (press for red, hold back for green) and Tower (planning). Some employers add numerical and logical games (Magnitudes, Sequences, Shapes, Letters), as candidates and prep providers report. All of them respond to practice within days, and much of the early gain is simply familiarity with the mechanic. This is where your preparation time belongs; the complete pymetrics games guide describes each game.

Situational judgement and simulations — partly preparable. You cannot rehearse the answers, but you can research the employer's stated values and think about what good judgement looks like in that context before you sit down cold.

What to practise

If your invitation includes cognitive games, practise the exact mechanics until nothing about them is new:

• Digit span — the Digits game, and backward digit span to stretch your span further.
• Flanker arrows — the centre-arrow half of Arrows, then task switching for changing rules mid-game.
• Reaction time — the speed side of Stop.
• N-back — the Letters game (2-back), if your invitation includes the extra games; numerosity for Magnitudes.
• Balloon risk — not to "win", only so the pumping mechanic is not new on the day.

Each run is scored on the server against fixed norm tables and returns a percentile, which is the useful part: you find your weakest category in about fifteen minutes and put the rest of your time there rather than into whichever game you like best.

Does a result follow you between employers?

This is the question worth asking your recruiter, and it is a real difference from one-off assessments.

For pymetrics, candidates and prep providers consistently report a 330-day rule: you can play the games once in that period, and the same result goes to every employer that invites you to them in that time, each judging it against its own role profile. Harver does not state the rule on its public site. That has two consequences: a good result saves you sitting the same games repeatedly, and a rushed one can shadow later applications.

Policies can vary by employer and region, and they change. Nobody online can tell you your situation reliably — the recruitment team that invited you can. The safe operating assumption either way: treat your first attempt as the one that counts everywhere.

Preparing whichever name is on the invite

• Read the invitation for the module list, not the brand. It tells you which of the categories above you actually face.
• Remove all unfamiliarity from the cognitive games. This is the largest controllable gain and it is available today.
• Play the behavioural tasks honestly and consistently. Pick a natural approach and hold it across every round.
• Fix the environment: quiet room, charged device, notifications off, stable connection. Timed tasks punish friction.
• Take it when you are sharp, inside the deadline window — not at 1 a.m. on the final day.

Fewer large graduate employers send the games than older lists suggest: JPMorgan, Goldman Sachs and Citi candidates no longer report them in 2026, and most BCG offices have moved to other tests. The employer table in our pymetrics guide gives the current picture with confidence labels.

For module-by-module detail see the Harver assessment guide and the complete pymetrics games guide. For the wider landscape, start from the complete guide to game-based assessments, and for what "passing" means when no vendor publishes a cut score, see what counts as a good game assessment score.

FAQ

What is the difference between Harver and pymetrics? Harver acquired pymetrics, so they are no longer competitors — the pymetrics-style games now sit within Harver's broader hiring platform alongside situational judgement and simulation modules. Which one your invitation names matters less than which modules it includes.

Is pymetrics still a separate product? No. Harver markets the games as its game-based behavioural assessments, run on the Harver platform with its other modules. Employers and candidates still call them the pymetrics games.

Which parts can I practise for? The cognitive games — memory, speed, attention, numerical fluency — respond well to practice. The behavioural and trait games do not, and attempting to game them usually produces an inconsistent profile.

Do I get the same games as every other candidate? Not necessarily. Employers select modules, and the combination differs by company and role. Your invitation is the only reliable description of what you will sit.

Can my result be reused by another employer? For the pymetrics games, reportedly yes: candidates describe one attempt per 330 days, shared with other employers using the games in that window. Harver does not publish the rule, so confirm with the recruitment team that invited you.

Is there a pass mark? No published one. Results are compared against other applicants and to a profile the employer defines for the role, so aim to be comfortably above a largely unprepared field rather than at a specific number.

GamePrep is an independent practice platform and is not affiliated with Harver, pymetrics or any assessment vendor. Vendor ownership, module names and data-reuse policies change — verify against your invitation and the employer's recruitment team. We provide related practice only — no leaked questions or answers.