Transition model
Train the ability, the item and the system
Most weak preparation stops at percentages, logic and patterns. Strong preparation also rehearses what the screen asks you to do and how the assessment behaves.
Ability
What mental operation is measured? Interpret data, derive a necessary conclusion, or infer a visual rule.
% · ratio · inference · patternItem
What must you do now? Select one option, rank every row, mark dates, or construct a full next state.
select → rank → schedule → constructSystem
How does the assessment behave? Fixed or adaptive, return allowed or locked, marks or percentile language.
fixed / adaptive · marks / percentileSystem difference
Same ability. Different response contract.
The arithmetic and disciplined inference transfer strongly. The output, navigation and score language do not transfer automatically.
Classical EPSO reasoning
- Ability families
- Verbal, numerical and abstract reasoning.
- Response
- Usually select one complete answer option.
- Architecture
- Typically a fixed set with visible section counts defined by the notice or invitation.
- Score language
- Marks and pass rules set by the notice.
SHL G+ / Interactive
- Ability families
- Numerical, deductive and inductive reasoning.
- Response
- Select in some variants; rank, schedule, adjust or construct in interactive variants.
- Architecture
- SHL describes G+ products as adaptive. Employer configuration and invitation remain decisive.
- Score language
- Overall and component ability scores, often reported against a comparison group. Percentile is not percentage correct.
Two delivery architectures
A fixed sequence adds marks. An adaptive sequence updates an ability estimate after each item.
Raw accuracy and percentile are different measures
The same performance can be shown as correct answers or as a position within a reference group.
Illustrative values only. A percentile does not mean percentage correct.
Activity-based response
Interactive means the answer is a configuration
The reasoning may end in one number or one rule. The response does not. You may need to rank every row, place every required event, adjust all parts of a chart, or construct a complete visual state. Solve first. Then encode and audit the whole configuration.
Interactive example 01 · Ranking
Rank every item, not only the winner
Actual and target results are North 150/120, East 176/160 and West 170/200. Rank all regions by percentage of target achieved. Rank 1 means highest.
Interactive example 02 · Scheduling
Build the only schedule that satisfies every rule
Place Mara (M), Lia (L) and Niko (N) from Monday to Friday. Mara is immediately before Lia; Mara is not Monday; Lia is not Thursday; Niko is Friday.
Interactive example 03 · Chart adjustment
Convert percentages into a complete chart state
Set quarterly output to 160 units. Standard must be 60% of the total and Premium 40%. Enter both quantities and set the split.
96 + 64 = 160 · 60% + 40% = 100%
Interactive example 04 · Zone construction
Construct every cell of the next state
The solid and hatched cells each move one corner clockwise. Paint the complete next 2 × 2 state.
These are original EPSOHQ teaching diagrams based on the SHL+ versus EPSO book. They illustrate transferable interaction mechanics. They are not screenshots or replicas of an official SHL interface, and they do not replace the tutorial for your assigned assessment.
Transfer labs
See the format shift side by side
These simplified original examples isolate the change in response method. The left side uses a familiar EPSO contract. The right side keeps the core skill but changes the action.
Lab 01 · Numerical
One calculation becomes a complete ranking
Which team exceeded its own target by the greatest percentage? Select one answer.
| Team | Target | Actual |
|---|---|---|
| Atlas | 400 | 440 |
| Boreal | 480 | 480 |
| Cedar | 280 | 240 |
Rank every team by actual ÷ personal target. Rank 1 means highest.
Lab 02 · Verbal → Deductive
One decisive sentence becomes a decisive chain
Does the conclusion follow from the passage? Stay inside the stated evidence.
Only files cleared by Legal may be released. File K was released.
Place Mara (M), Lia (L) and Niko (N) into the only schedule that satisfies every rule.
M-LMara immediately before Lia
L < NLia before Niko
M ≠ 1Mara not Monday
L ≠ 4Lia not Thursday
N = 5Niko on Friday
Lab 03 · Abstract → Inductive
Recognising a figure becomes constructing its state
The arrow rotates 90° clockwise. Select the complete next figure.
Both marked cells move one corner clockwise. Predict every cell, then paint the next state.
Deep practice
Four complete questions with full explanations
Solve the left pane first. Then follow the trace on the right. Each explanation identifies the correct answer, the proof or calculation, why the distractors fail, and the final response check.
Solve from evidence.
Enter the intended state.
Touch every required row or zone.
Check highest, lowest, before or after.
Commit and close the item.
Exhibit A · Numerical percentage
Percentage decrease: identify the old value as the base
Full question
| Region | 2022 | 2023 |
|---|---|---|
| North | 145 | 160 |
| South | 210 | 185 |
| East | 95 | 115 |
What was the percentage decrease in South region sales from 2022 to 2023?
Full explanation
- State the output. The question asks for a percentage decrease, not the absolute fall.
- Find the change. Sales fell from 210 to 185, so the decrease is 25.210 − 185 = 25
- Use the starting value as base. “From 2022” makes 210 the denominator.25 ÷ 210 × 100 = 11.904…%
- Round only at the end. 11.904…% rounds to 11.9%, matching option B.
Exhibit B · Interactive threshold
Two-condition threshold: select every qualifying row
Full question
| Person | Actual | Target | Rating |
|---|---|---|---|
| Ava | 440 | 400 | 4.3 |
| Ben | 480 | 480 | 4.6 |
| Cara | 300 | 280 | 4.0 |
| Dario | 360 | 320 | 4.5 |
A bonus requires performance of at least 105% of personal target AND a rating of at least 4.2. Select every eligible person.
Full explanation
- Translate both thresholds. “At least” includes equality. Both conditions must hold for the same person.actual ÷ target ≥ 1.05 AND rating ≥ 4.2
- Ava. 440 ÷ 400 = 1.10; 4.3 ≥ 4.2. Both conditions pass.
- Ben / Cara. Ben fails performance at 100%. Cara passes performance at 107.1% but fails the rating condition at 4.0.
- Dario. 360 ÷ 320 = 1.125; 4.5 ≥ 4.2. Both conditions pass.
Exhibit C · Deductive necessity
Must be true: prove the chain, not the plausible story
Full question
If Project A is approved, Project B will not be completed on time. Project C can start only if Project B is completed on time. Project C has started.
Which statement must be true?
Full explanation
- Translate “only if”. C can start only if B is on time means C → B on time.
- Apply the stated fact. C has started, so B must have been completed on time.
- Use contradiction. If A had been approved, B would not be on time. But B is on time. Therefore A was not approved.
Exhibit D · Inductive construction
Construct the next state: track both moving cells
Full question
The solid and hatched cells each move one corner clockwise while remaining opposite. Which option is the full next state?
Full explanation
| Variable | Observed rule | Next state |
|---|---|---|
| Solid cell | BL → TL → TR → BR | Bottom-right |
| Hatched cell | TR → BR → BL → TL | Top-left |
| Relationship | Always opposite | Preserved |
- Separate variables. Track the solid and hatched cells independently. Do not describe them as one vague pair.
- Advance each path once. Solid moves top-right → bottom-right. Hatched moves bottom-left → top-left.
- Validate the relationship. The predicted cells remain diagonally opposite, so the rule still explains the full state.
Adaptive execution
Reset in three words. Stop on lost progress.
A 24-in-36 practice run averages 90 seconds per item. That average is a portfolio target, not an entitlement for every question.
Numerical, deductive or inductive?
Number, rank, dates, placement or next state?
Unit, threshold, logic rule or transformation?
Can you state task, output and constraints?
Has the solution space narrowed materially?
Validate direction once. Submit and close.
Replace “careless” with an error code
A useful review says where execution failed. One label tells you what to drill next.
N-BASEMethod
Wrong base, reversed conditional or missed visual variable.
I-ENCODEInteraction
Reasoning correct; rank, date, zone or default state entered wrongly.
S-PACESystem
Time debt, return assumption or adaptive-score speculation.
E-RUSHState
Rush, freeze or carry-over from the previous item.
Book programme
Fourteen-day transition programme
Move from method rebuild to interface control, then taper. Record accuracy, time, confidence and error code after every set.
Run untimed and timed diagnostics. Separate ability errors from interaction and pacing errors.
Drill ASK–BASE–UNIT–FORMULA, constraint notation and the visual variable ledger.
Practise rankings, schedules and constructed states. Add the full-response audit.
Switch families under time. Use Family–Output–Constraint before every item.
Apply the two gates. Review confidence errors and repeated codes, not only wrong answers.
Take one final realistic run, rehearse on the test-day device, then reduce volume and protect sleep.
Public source checks
EPSO's official testing page defines its verbal, numerical and abstract reasoning samples. SHL's official Verify Interactive G+ fact sheet describes numerical, deductive and inductive ability scores, adaptive activity-based delivery, a 36-minute allowance and a maximum of 24 questions. Your invitation remains final.