Verbal Reasoning Test 2026: How to Pass Cannot Say

The rule that decides a verbal reasoning score

Work each statement through three questions, in this order.

1. Does the passage state it, or force it? If the passage says it outright, or if the passage cannot be accurate while the statement is wrong, mark it true.
2. Does the passage rule it out? If the passage says something that cannot hold at the same time as the statement, mark it false.
3. Anything else is cannot say. Including statements that are almost certainly true in the real world.

Step three is where scores are lost. Cannot say is the correct answer whenever the passage neither forces nor forbids the statement, whatever your confidence about the outside world. Candidates who treat it as a hedge for uncertainty give away real trues and falses.

SHL, whose formats sit behind a large share of graduate screens, words the instruction this way on its own candidate site: you are asked to "determine whether each statement is true or false, or whether you cannot say, given the information in the passage." The last five words carry the whole test. That page is at shl.com/shldirect, checked 29 September 2026.

What the verbal reasoning test actually measures

SHL's own candidate report describes the verbal test as measuring "the ability to evaluate the logic of various kinds of arguments," and adds that it "emphasises understanding, using and evaluating verbal information rather than language usage, spelling or grammar."

That sentence should change how you prepare. Vocabulary work, wider reading, and grammar revision do nothing for this score. The skill being measured is narrow: holding a claim against a text and judging what the text establishes. It improves with repetition on the format, and with very little else.

The same report adds that the passages and tasks are "relevant to a business environment," and that people who do well "tend to have the capacity to readily understand reports and documents." That is the reason firms screen with it. A consultant reads a client document, and the job is to say what it supports and where it stops, at speed, in front of someone who will check.

It also explains why subject knowledge is irrelevant. You are not expected to know anything about the topic before you read it, and knowing something is a liability rather than an advantage, because familiarity is what tempts you to fill a gap the passage left open.

Worked passage one: the real world is not evidence

The firm opened offices in Lisbon and Warsaw in 2024. Both offices recruit analysts directly from local universities. The Lisbon office reached forty staff within its first year.

"The firm opened at least two offices in 2024." True. Two are named, and two satisfies at least two. Nothing in the passage limits the total to two, and nothing needs to.

"The Warsaw office reached forty staff within its first year." Cannot say. The passage gives that figure for Lisbon only. Both offices opened in the same year and both recruit the same way, so the statement is plausible. The test does not ask about plausibility.

"The Lisbon office recruits only from local universities." Cannot say. The passage says both offices recruit directly from local universities. It says nothing about whether they recruit anywhere else. The word "only" is doing all the work, and the text never licenses it.

That last item is the most common single trap in the format. A statement that would be true without one qualifier becomes unsupported once the qualifier is added, and a fast reader marks it true because the rest of the sentence matched. Underline every absolute word in the statement before you look back at the passage.

Worked passage two: the converse is not the claim

Every candidate who applies before the deadline receives the online assessment. The firm also sends it to candidates referred by an employee after the deadline has passed.

"Some candidates receive the assessment after the deadline has passed." True. The second sentence states exactly that for referred candidates.

"Every candidate who receives the assessment applied before the deadline." False. Referred candidates receive it after the deadline, so at least one recipient did not apply before it. The passage rules the statement out, which is what makes this a false rather than a cannot say.

The second statement reverses the direction of the first sentence. "Everyone who applies in time gets it" establishes nothing on its own about who else might get it. Reversing a conditional and treating the result as settled is the error that separates one band from another on this format, and it stays invisible unless you check the direction of each claim deliberately.

A quick test that catches it: rewrite the statement as "if A then B," rewrite the passage the same way, and compare the order of the letters. If they are swapped, the statement needs its own support somewhere in the text.

Worked passage three: false needs a contradiction

All of the firm's offices run a written case at final round. The London office also runs a group exercise at final round. No office other than London runs a group exercise.

"Every office runs a group exercise." False. The third sentence contradicts it directly.

"A candidate interviewing in London does both a written case and a group exercise at final round." True. The first sentence covers every office, and the second adds the group exercise for London.

"The written case is the same in every office." Cannot say. Every office runs one. Nothing in the passage describes the content, so sameness is neither established nor excluded.

Compare this passage with the first one. Here a false answer is available because the text contains an explicit exclusion. Strip the third sentence and "every office runs a group exercise" becomes cannot say, because a passage that mentions London does not thereby deny the rest of the network. Reach for false only when you can point at the words that forbid the statement, and treat the absence of those words as a cannot say.

Where scores are lost

Four failure modes account for most of the damage, in rough order of cost.

• Importing outside knowledge. You know the sector, so you fill a gap the passage left. Every gap the passage leaves is a cannot say.
• Reversing a conditional. Covered above, and the hardest to catch at speed.
• Ignoring absolute words. All, only, always, never, every, none. Underline them in the statement, then look in the passage for the words that license them.
• Using cannot say as a confidence setting. Some candidates mark it whenever they feel unsure, which converts trues and falses into a third of a mark each. Decide by the rule and let your comfort follow.

There is a fifth, quieter one worth naming: rereading. A candidate who skims the passage once badly then rereads it for every statement runs out of clock on a test that most people finish. Read it properly once, then work each statement against what you already hold.

Write down which of the five caused each item you got wrong. After twenty items one of them will be recurring, and that one is your whole preparation.

Timing on a verbal reasoning test, and where to spend it

These tests run under a fixed clock and are built so that most candidates finish. SHL describes its cognitive assessments at shl.com/products/assessments as "automated, remotely proctored, and ask fewer questions thanks to our unique partial-scoring methodology." Fewer items means each one carries more weight, so a careless answer costs more here than on a long test.

That changes the allocation. Read the passage once, properly, before you look at any statement. Then treat each statement as a separate exercise against a text you have already absorbed. Most of the time saved in this format comes from not going back.

One rule about skipping is publisher specific and worth knowing. TestGorilla, which some firms use for their screens, tells candidates plainly: "Once skipped, you cannot go back and answer a question. We do not recommend skipping questions. Skipped questions count as wrong answers. If you don't know the answer, guess." Its candidate guide is at candidates.testgorilla.com, checked 29 September 2026. On that platform a blank item and a wrong item score the same, so never leave one empty.

A final mechanical point. SHL's report notes that its results stay on file for a period, which means the sitting is usually a one time event inside a cycle. Treat the clock as the constraint it is, and decide in advance how long you will spend on a statement before you commit and move on.

How it is scored, and what nobody publishes

SHL's candidate report shows two things for each test: a percentile against a named comparison group, and a letter grade on a five point scale printed as E, D, C, B, A. The comparison group appears on the report itself. On the sample report SHL publishes, the verbal and numerical results are compared with a "General Population 2006" comparison group, so the benchmark is a specific and dated population rather than an absolute standard.

The report also states that "test results remain valid for about 12 to 18 months." That is the most useful line in it. A result that carries for over a year makes a second sitting inside the same recruiting cycle unlikely to be offered, so the first one is the one that counts.

What nobody publishes is the cutoff. SHL reports a percentile and a band, and the decision about which band continues belongs to the employer. Employers do not publish that decision, and no publisher can, because the same test serves different roles at different bars. Any page that gives you the score you need for a named firm has invented it.

Two more things are outside anyone's reach. Which publisher your test comes from varies by firm, by office, and by cycle, and none of them list it. The composition of your sitting varies too, because publishers assemble an assessment from modules the employer selects. Your invitation email names the platform, and it is the only source on this that is current by definition.

How to practice this

Do items, read the correction on everything you got wrong, and label the failure mode. The label matters more than the score, because the four modes take different corrections and a raw percentage hides which one you are repeating.

SHL publishes example questions and a full length practice test for candidates on its own site, covering verbal, numerical, inductive, calculation, checking, personality, motivation, and what it labels "situational judgement". It is the publisher's own material, it is free, and it is the first place to go. Start there rather than with a paid question bank, then use the bank only if you exhaust it.

We do not have a verbal drill, and saying otherwise would be easy and false. What we do have is the same discipline applied to exhibits inside a case, where the failure mode is identical: a claim asserted past what the data supports. The free drills score that, and how an answer is assessed sets out the weights they use, read from the code that does the scoring. Our assessment simulators cover the firm built formats rather than publisher tests, and none of the six is live yet, which that page says itself.

For the publisher specific mechanics, SHL style assessments covers the adaptive behavior and the personality section. For the other half of most screens, numerical reasoning covers where the time actually goes on exhibit questions.