An online assessment is the set of timed tests and questionnaires an employer sends you after you apply, usually before any interview. Most of it is a psychometric test in one form or another: numerical, verbal and abstract reasoning, a personality questionnaire, a situational judgement test (SJT) or a short series of games. A typical graduate battery takes 30 to 70 minutes in the browser. This guide shows every format seen in real 2026 graduate rounds, which vendors deliver them, how they are scored and how to prepare in two weeks.
Key facts
- When
- after the application form or CV, before interviews
- Time
- 30–70 min in total; single tests 4–36 min
- Sections
- numerical, verbal, inductive, deductive, personality, SJT, games
- Publishers
- SHL, Aon, Korn Ferry, Cappfinity, Arctic Shores, eligo, cyquest
- Answer format
- multiple choice, true/false/cannot say, ranking, sliders, games
- Scoring
- percentile against a norm group; employer sets the cut-off
- Deadline
- often a few days, stated in the invitation
What is an online assessment?
You get an e-mail with a link, open the test platform in your browser and work through several modules, in one sitting or several. The platform belongs to a test publisher (SHL, Aon, eligo…), but the employer chooses which modules you see, in which language, and where the pass mark sits.
The tests fall into three families:
- Ability tests have right and wrong answers and a clock: numerical, verbal, abstract (inductive), deductive and critical reasoning, number series, speed and attention tasks.
- Behavioural questionnaires have no right answers in the usual sense: personality questionnaires, work-style sliders and values rankings.
- Judgement and simulation tasks sit in between: situational judgement tests, planning or in-tray tasks, game-based tasks and one-way video interviews.
"Psychometric" simply means the test is built and scored with statistical methods, so your result can be compared with a reference group.
The stages of a graduate application
The online assessment is the second filter, after the application form. The order is stable across employers:
| Stage | What happens |
|---|---|
| 1. Application | CV, form, sometimes an eligibility screen |
| 2. Online assessment | Ability tests, personality, SJT or games, from one vendor or a mix |
| 3. Video interview | One-way recorded answers or a live video call |
| 4. Interviews and cases | Personal fit and case interviews, sometimes several on one day |
| 5. Assessment centre | Presentations, role plays, group tasks; on site, virtual or hybrid |
| 6. Offer | Often within days of the final stage, sometimes on the same day |
Consultancies tend to go from the online test straight to case interviews; trainee programmes add a video interview and an assessment centre. See the assessment centre guide, the video interview guide and the case interview guide for the later stages.
Every psychometric test type, one by one
Each section below gives the format in a few lines and links to the full guide with original practice questions.
Numerical reasoning
You answer questions from tables, charts and short word problems: percentages, ratios, growth, unit rates. Formats vary more than the name suggests. SHL uses five-option multiple choice, TestGorilla three or four options, Cappfinity sometimes asks you to type an exact answer, and eligo and cyquest show one chart with a statement that you mark true, false or "cannot be determined". Traps: units in the chart title, figures below the fold, hover-only values, truncated axes. Full guide: numerical reasoning test.
Verbal reasoning
A short business text and a series of statements; for each one you decide true, false or cannot say based only on the text. Expect 12 to 18 statements in 6 to 9 minutes. The two layouts seen in 2026 were a scrolling passage (cyquest "Textanalyse", buttons RICHTIG / FALSCH / NICHT ABLEITBAR) and a brief split across six tabs (Aon, "wahr / falsch / unbestimmt"). Full guide: verbal reasoning test.
Inductive and abstract reasoning
Shapes follow hidden rules; you find the rule and apply it. Four quite different formats appeared in 2026: Aon's "pick the two of four grids that follow the rule" (about 6 seconds per item), eligo's 4×4 matrices with six answer tiles, cyquest's "Culture-Fair-Test" with options A–H, and Arctic Shores' puzzle where you build the next frame yourself. SHL's interactive version has you colour in a figure. Full guide: abstract reasoning tests.
Deductive reasoning
Rules or premises, and the question of what must follow. It appears as syllogisms, "which statement must be true" items, eligo's passage-plus-conclusion format and SHL's interactive scheduling tasks where you drag items into a timetable that satisfies every rule. Full guide: deductive reasoning test.
Critical thinking
Watson-Glaser-style items: recognise assumptions, judge inferences, evaluate arguments, draw conclusions. Full guide: critical thinking test.
Number series
Continue a sequence of numbers by finding the rule. cyquest's "Fortführen von Jahreszahlen" gave 16 series in 8 minutes with free number entry. It is most common in German-language tests (guide in German): Zahlenreihen fortsetzen.
Speed and working memory
Short, heavily timed blocks that test how fast you hold and transform information. In eligo's "select the larger / smaller value" task, nine symbols stand for the digits 1–9, the key is reshuffled on every trial, and you compare single values, sums, products or differences in blocks of 30 to 50 seconds. Full guide: working memory test.
Attention to detail and multitasking
Compare two versions of an address or record and decide same or different, sometimes while routing incoming messages at the same time. Full guide: attention to detail test.
In-tray and planning tasks
A simulated working day: prioritise messages or drag appointments into a plan while following a set of rules. cyquest's "Planungsaufgabe" used a count-up clock, so speed counted as well as correctness. Full guide: in-tray exercise.
Situational judgement tests (SJT)
Realistic work scenarios with responses to judge. In 2026, an employer-branded culture-match SJT (Korn Ferry) asked for the most and least effective of three options in 18 linked scenarios, untimed; a Cappfinity values ranking asked you to rank five options from 1 to 5 in 10 scenarios. Full guide: situational judgement test.
Personality questionnaires
How you prefer to work, not what you can do. Three formats dominate: forced-choice triads (SHL OPQ32r, 104 blocks, rank "most like me" and then the better of the remaining two), bipolar sliders (Cappfinity's 30-item work-style questionnaire, cyquest's 0–100 sliders) and agreement scales. Usually untimed, about 5 to 25 minutes. A personality assessment test cannot be "passed", but your profile is compared with the role. Full guide: personality test for the workplace.
Game-based assessments
Short interactive tasks that record speed, accuracy, hesitation, risk-taking and learning. Arctic Shores lets the employer choose up to nine tasks; one 2026 battery used three in about 32 minutes. Pymetrics, Aon and McKinsey Solve run their own game formats. Full guide: game-based assessment.
Emotion recognition
A face appears and you choose one of seven labels (six basic emotions plus neutral) in about four seconds. This is Arctic Shores' "Expressions" game. Full guide: emotion recognition test.
One-way video interviews
Not a psychometric test, but often part of the same invitation or the next step. You see a question, get a short thinking countdown and record an answer of 60 to 120 seconds. One 2026 version had nine questions and switched from English to German partway through. Full guide: HireVue and one-way video interviews.
Online assessment platforms and vendors compared
Employers rarely build their own tests. They license an online assessment platform and pick modules. Knowing the vendor tells you most of what to expect, so check the URL and footer of your invitation.
| Vendor / platform | Formats you will meet | Guide |
|---|---|---|
| SHL (TalentCentral) | General Ability (30 items, 36 min, 5 options, no back button), interactive General Ability (drag, tap, colour), OPQ32r personality | SHL test |
| Aon (cut-e) (smartPlayer) | scales clx inductive grids, scales verbal and numerical with tabbed briefs, short challenges | Aon test |
| Korn Ferry | Employer-branded culture-match SJTs; Talent Q Elements adaptive ability tests | Korn Ferry assessment |
| eligo (PERLS DUO) | Logic from text, chart analysis, 4×4 matrices, symbol speed task; per-item timers | eligo test |
| cyquest | Number series, charts, text analysis, Culture-Fair matrices, multitasking, planning task, personality sliders | cyquest (German) |
| Arctic Shores | Task-based games: puzzles, prediction from feedback, emotion recognition and more | Arctic Shores assessment |
| Cappfinity | Situational strengths and values ranking, work-style sliders, time-recorded numerical tests | Cappfinity |
| TestGorilla | Short skills tests (10–12 min each) from a large library: critical thinking, numerical, data literacy | TestGorilla |
| Applied (beapplied) | Anonymised written questions plus timed skill tests, e.g. an 18-item numerical test with hover charts | Applied |
Other names you may see in invitations, each with its own guide: HireVue, pymetrics / Harver, Watson-Glaser, Saville, Talogy, Sova, Criteria (CCAT), Wonderlic, Predictive Index, Thomas International, Hogan, Mercer Mettl, McKinsey Solve, BCG online case, Clevry, Alva Labs, Matrigma, Bryq, Plum, Revelian and IBM Kenexa. Their formats overlap with the ones above; for immersive day-in-the-life tests see job simulation assessment.
What a typical battery looks like: six real examples
A battery is the combination of modules one employer sends. These are batteries observed in autumn 2026 graduate rounds, labelled by vendor. Employers change them by role, country and year, so treat them as examples, not promises.
eligo PERLS DUO battery (English): four modules, per-item timers
In a 2026 application round, an eligo PERLS DUO battery in English had four modules with three untimed sample tasks and 14 scored items each:
| Module | Format | Time |
|---|---|---|
| Logical reasoning | Passage + conclusion: correct / indeterminate / false | 1:30 per item |
| Chart reasoning | Chart + statement: true / indeterminate / false; magnifier for zoom | 1:00 per item |
| Figural matrices | 4×4 matrix, six answer tiles, two or three layered rules | 1:00 per item |
| Processing speed | Symbols for digits 1–9, pick the larger or smaller value | 12 blocks of 0:30–0:50 |
Plan for about an hour including instructions. The result report gave a percentile per module in low, medium and high bands. A German-language eligo battery used the same matrices and chart items. Details: eligo test.
cyquest battery (German, 7 sections)
This battery was in German and used the informal "du":
| Part | Format | Items · time |
|---|---|---|
| Fortführen von Jahreszahlen | Continue year sequences, typed answer | 16 · 8:00 |
| Analyse von Diagrammen | RICHTIG / FALSCH / NICHT ABLEITBAR | 12 · about 10:00 |
| Textanalyse | Two scrolling passages, same three buttons | 12 · about 9:00 |
| Culture-Fair-Test | 4×4 figure matrices, options A–H | 14 · 10:00 |
| Multitaskingtest | Address pairs same/different plus message routing | 17 screens × 6 rows · about 6:00 |
| Planungsaufgabe | Drag-and-drop room schedule, 7 rules, count-up clock | 1 scenario |
| Persönlichkeitstest | 0–100 sliders | 20 pages × 5 statements, untimed |
The timed parts alone add up to about 45 minutes, before the planning task and the questionnaire. Details: cyquest (German).
Aon and Korn Ferry battery: two fast tests and an SJT
| Part | Vendor | Format | Items · time |
|---|---|---|---|
| Inductive reasoning (scales clx) | Aon | Pick the 2 of 4 grids that follow the rule of two example grids | 3 practice, then 57 items reached in 6:00 |
| Verbal reasoning | Aon | wahr / falsch / unbestimmt; facts split over six tabs | 3 practice + 18 · 6:00 |
| Employer-branded culture-match SJT | Korn Ferry | Most and least effective of 3 options, one storyline | 1 + 18 · untimed (about 4 min) |
The two Aon tests are short but brutally paced: about 6 seconds per inductive item and 20 seconds per verbal statement. Details: Aon test, Korn Ferry assessment.
TestGorilla battery (English): three short tests
| Test | Format | Items · time |
|---|---|---|
| Critical thinking | Single and multiple choice | 4 practice + 16 · 12:00 |
| Numerical reasoning | Word problems and charts, "Select only one answer" | 4 practice + 12 · 10:00 |
| Data literacy | Charts including truncated axes, some "select all that apply" | 4 practice + 12 · 10:00 |
Navigation was forward only. Details: TestGorilla.
Cappfinity battery (English): time recorded, not limited
| Part | Format | Items |
|---|---|---|
| Values ranking | Rank five responses 1–5 per scenario | 10 |
| Work-style questionnaire | Bipolar sliders | 30 |
| Numerical reasoning | Pick one of five or type the answer; a stopwatch counts up per item | at least 12 |
A calculator was permitted, and the intro asked you to allow at least 20 minutes for the numerical part. Details: Cappfinity.
Arctic Shores battery (German): three games
| Game | What you do | Size |
|---|---|---|
| Rätsel (Order) | Build the fourth frame from shape, pattern, rotation, size and position rules | 1 practice + 17 |
| Nacht oder Tag (Predict) | Learn from feedback which symbol patterns predict night or day | about 90 rounds |
| Gesichtsausdruck (Expressions) | Pick one of seven emotion labels for a CGI face | 7 practice + 60, about 4 s each |
The three games took about 32 minutes in total. The timer was a filling stopwatch icon without seconds. Details: Arctic Shores assessment.
One more pattern: SHL ability + personality
One 2026 round paired SHL General Ability (30 items, 36:00, no back button) with the OPQ32r straight after it (104 triads, untimed, about 18–20 minutes). Another used the interactive General Ability test instead: about 23 drag-and-tap tasks within one 36-minute countdown. See SHL test.
Scoring, norm groups and the psychometric test conversion table
Raw score, norm group, percentile
Ability tests start with a raw score, usually the number of correct answers. On its own that number means little, because 14 out of 30 might be excellent on a hard test and weak on an easy one. So the vendor compares your raw score with a norm group: a large sample of people who took the same test, ideally a similar population (graduates, applicants for professional roles). The result is a percentile: "percentile 70" means you scored higher than 70 % of the norm group.
One eligo report, for example, showed one percentile per module against a large comparison group, grouped into low, medium and high bands. Arctic Shores feedback, where the employer releases it, shows each behaviour on a 1–10 scale against a norm group.
Adaptive tests
Some tests choose the next item based on your previous answers. SHL describes Verify G+ as adaptive, and Korn Ferry's Talent Q Elements are adaptive by design. In an adaptive test your score depends on the difficulty of the items you got right, not only on how many, so do not panic if items feel hard: that often means you are doing well. None of the eligo, cyquest, Aon, TestGorilla or Applied tests observed in 2026 showed signs of adaptivity; they had fixed item counts.
Psychometric test conversion table
Reports often use standard scales instead of percentiles. The common ones assume a normal distribution:
| Sten (1–10) | T-score | Percentile range | Reads as |
|---|---|---|---|
| 1 | below 30 | 1–2 | well below average |
| 2 | 30–35 | 3–7 | |
| 3 | 35–40 | 8–16 | below average |
| 4 | 40–45 | 17–31 | |
| 5 | 45–50 | 32–50 | average |
| 6 | 50–55 | 51–69 | average |
| 7 | 55–60 | 70–84 | |
| 8 | 60–65 | 85–93 | above average |
| 9 | 65–70 | 94–97 | |
| 10 | above 70 | 98–99 | well above average |
A sten score has a mean of 5.5 and a standard deviation of 2; a T-score has a mean of 50 and a standard deviation of 10. The middle band (sten 5–6, roughly percentiles 31–69) covers about 38 % of people.
Who decides whether you pass
The vendor provides the norm; the employer sets the cut-off, and it may differ by role. Some employers use a hard hurdle per test, others combine the tests into one score, and some combine the test with your CV. No universal pass mark exists, and none was shown in the 2026 tests.
Personality and SJT scoring
Personality questionnaires are compared with a norm group and with the profile the role needs. Many include a consistency indicator. SJTs are scored against a key; ranking and rating formats often give partial credit for answers close to it. Understand the format, know the employer's values and answer honestly and consistently. See personality tests and SJTs.
Time management: know your seconds per item
The clock, not the difficulty, decides most ability tests. Work out your budget before you start:
| Test | Items · time | Seconds per item |
|---|---|---|
| Aon scales clx | 57 reached · 6:00 | about 6 |
| Arctic Shores Expressions | 60 faces | about 4 |
| Aon verbal | 18 · 6:00 | 20 |
| cyquest number series | 16 · 8:00 | 30 |
| cyquest Culture-Fair-Test | 14 · 10:00 | about 43 |
| TestGorilla critical thinking | 16 · 12:00 | 45 |
| cyquest charts | 12 · about 10:00 | about 50 |
| TestGorilla numerical | 12 · 10:00 | 50 |
| eligo charts and matrices | per item | 60 |
| SHL General Ability | 30 · 36:00 | 72 |
| eligo logic | per item | 90 |
| Applied numerical | 18 · 30:00 | 100 |
Rules that hold across vendors:
- Block timer versus item timer. With one countdown for the block (SHL, Aon, cyquest, TestGorilla), you can borrow time from easy items; set a checkpoint at the halfway mark. With a timer per item (eligo), time cannot be carried over, so commit and move on.
- No back button means no "come back later". SHL General Ability and TestGorilla are forward only. Answer, mark your best guess, and go on.
- Skip the sink. If an item will clearly take twice your budget, pick the most likely answer and move on.
- Practice items are free time. Use untimed samples to learn the interface: magnifier, tabs, options below the fold.
- Time recorded is still time. Cappfinity's numerical test and the cyquest planning task count your time even without a countdown.
- Answer everything. Unless the instructions state a penalty for wrong answers, a blank scores the same as a wrong answer.
Setup, devices and accessibility adjustments
Device and environment
- Use a laptop or desktop with a mouse. Some platforms run on phones (SHL's interactive test is mobile-ready, and Aon lists phones and tablets as allowed), but fast clicking, dragging and scrolling are easier with a mouse and a large screen. cyquest recommends a laptop or PC.
- Check the screen layout in the practice items. In 2026 tests, answer tiles sat below the fold on eligo, Aon verbal hid facts on six tabs and at the bottom of scrolling panels, and Applied charts showed exact values only on hover. eligo's chart magnifier hid the introduction sentence while zoomed.
- Stable connection, current browser, no extensions. Close other tabs and mute notifications. Cappfinity's instructions warn against browser extensions that alter the page.
- Paper, pen and, where allowed, a calculator. One Cappfinity numerical test stated that a calculator was permitted; chart and logic items go faster with notes. For Aon clx and game-based tasks you will not need them.
- Block the full time. Assume a timed test cannot be paused once it starts. Plan the stated duration plus 15 minutes.
- Read the rules of the platform. Tests ask you to work alone and without outside help, and some are proctored (remote supervision with an ID check, for example). Take that at face value: the test only helps you if it measures you.
Accessibility adjustments
If you have a disability, a learning difficulty such as dyslexia, colour-vision deficiency or another condition that affects test performance, ask the employer's recruiting team about adjustments before you start: extra time, an untimed or alternative version, screen-reader-compatible formats or a different test are set up in advance and cannot be added once a timed test is running. Points to raise:
- Matrices and SHL's interactive items rely on colour and shape; tell the employer about colour-vision needs.
- Speed tasks (eligo processing speed, Arctic Shores Expressions at about 4 seconds per face) are hard to adapt on the fly, so ask early.
- Verbal tests with tabbed or scrolling texts can be slow with screen magnifiers; ask whether an alternative layout exists.
A 2-week prep plan
Two weeks of about an hour a day is enough for most graduate batteries. Adapt the plan once you know your vendor.
| Day | Focus | What to do |
|---|---|---|
| 1 | Diagnosis | Identify the vendor from the invitation. Read the vendor guide. Do one timed practice set per test type in your battery; note your seconds per item |
| 2 | Numerical I | Percentages, percentage change, ratios, unit conversions. 20 items, no calculator for the easy steps |
| 3 | Verbal | True / false / cannot say: the decision rule, then 30 statements at your target pace |
| 4 | Inductive | The rule checklist (count, position, rotation, colour, size, movement). 20 matrices or grids |
| 5 | Numerical II | Chart and table items with traps: units, bases, truncated axes, data below the fold |
| 6 | Deductive and critical thinking | Syllogisms, "must be true", assumptions; sketch case tables |
| 7 | Rest or light review | Re-read your error log; no new material |
| 8 | Vendor format | Use the vendor's free practice (SHL, Cappfinity and cyquest offer practice; TestGorilla and Aon open with practice items) or the matching guide |
| 9 | Speed and attention | Symbol coding, mental arithmetic to 9 × 9, same/different drills in 2-minute blocks |
| 10 | Personality and SJT | Read the employer's values; list your honest trade-offs (team vs alone, plan vs improvise); practise the format, not "ideal" answers |
| 11 | Full mock | One complete timed battery in the real order, in one sitting |
| 12 | Fix weak spots | Re-do the item types that cost the most time in the mock |
| 13 | Setup check | Device, browser, connection, room, paper, calculator rules; read the invitation again |
| 14 | Test day | Short warm-up (5 easy items), then take the test rested |
Keep an error log: for every wrong or slow item, write down the trap in one line ("percentage of the wrong base", "answered cannot say with the wrong tab open"). By day 12 the list shows you exactly what to fix.
If you only have three days, do day 1, one session on your weakest test type, and the full mock.
Sources
- SHL, Verify G+ fact sheet (30 questions, 36 minutes, adaptive)
- SHL, OPQ product fact sheet (104 questions, untimed, 15–25 minutes).pdf)
- SHL candidate practice tests
- Aon, Candidate Guide on Aon Assessments (devices, browser, verbal and numerical timing)
- GraduatesFirst, Korn Ferry Talent Q (Elements adaptive ability tests)
- Arctic Shores, candidate guidance (up to nine tasks, setup advice, 1–10 feedback scale against a norm group)
- Cappfinity Preparation Hub
- TestGorilla test library, cognitive ability tests
- The Interview Guys, Arctic Shores test guide (retakes set by the employer)
Related guides
- Test types: numerical · verbal · abstract / inductive · deductive · critical thinking · number series (German) · working memory · attention to detail · in-tray and planning · SJT · personality · game-based · emotion recognition · video interview · assessment centre · case interview
- Vendors: SHL · Aon · Korn Ferry · eligo · cyquest (German) · Arctic Shores · Cappfinity · TestGorilla · Applied
- German version: Online-Assessment: der komplette Guide

MSc Psicologia, 7 anni di test fatti.
Ultimo aggiornamento: Oct 1, 2026