Pearson TalentLens 2026: Raven's APM, Watson-Glaser & BMCT guide

Updated April 2026 · 13 min read · Pearson assessment division

Pearson TalentLens delivers the most academically respected cognitive tests in commercial use. Raven's Progressive Matrices was developed by John C. Raven in 1936 and has been used in psychometric research for ninety years. The Watson-Glaser Critical Thinking Appraisal dates from 1925 and remains the standard test for legal and policy roles. These are not gimmicky modern assessments — they're the foundational tests psychology and HR have validated across decades.

Raven's Advanced Progressive Matrices (APM)

The classic abstract reasoning test. You see a 3x3 grid of patterns with one cell missing, and select which option (out of eight) completes the pattern based on the rules visible in the other eight cells. Rules combine: rotation, addition/subtraction of elements, alignment, colour-shift, and so on. Difficulty increases through the test.

The classic APM has 36 items in two sets. Set I (12 questions) is a warm-up; Set II (36 questions) is the scored portion. Time limit: 40 minutes for the standard version (untimed for clinical use). Most commercial deployments use the timed format.

Raven's APM-III (newer, adaptive)

Pearson released APM-III in 2018 as an adaptive replacement for the classic APM. Same matrix format, but each candidate sees a calibrated subset of items based on their performance. The adaptive version is faster (typically 20-25 minutes) and more secure (different candidates see different items).

Watson-Glaser Critical Thinking Appraisal

The standard test for roles requiring legal reasoning, policy analysis, or evidence evaluation. Five sub-sections:

Watson-Glaser is heavy with subtle distinctions: "True / Probably True / Insufficient Data / Probably False / False" rather than the simpler 3-option True/False/Cannot Say used by SHL. Used at Linklaters, Allen & Overy, Clifford Chance, Slaughter and May, Hogan Lovells for trainee solicitor screening. Also used by UK Civil Service for senior policy roles.

Bennett Mechanical Comprehension Test (BMCT-II)

The standard mechanical reasoning test for engineering and skilled-trade roles. Tests understanding of levers, pulleys, gears, springs, fluid dynamics, electrical circuits, and force vectors. 55 questions in 25 minutes (some versions: 30 questions in 30 minutes).

Used heavily in aerospace (BAE Systems, Airbus, Boeing) and oil & gas (Shell, BP, Equinor). Also used for technician roles at Network Rail, National Grid, and the Royal Navy / RAF aptitude batteries.

Adaptive Matrices

Pearson's modern adaptive abstract reasoning test, distinct from APM-III. Designed for shorter time windows (15 minutes) and very high-volume screening. Used by some graduate schemes as a lightweight first-stage cognitive filter.

Companies using Pearson TalentLens

UK Civil Service Fast Stream uses Watson-Glaser for analytical and policy roles. MI5 / MI6 / GCHQ use Raven's variants in their cognitive batteries. BP, Shell, and ScottishPower use BMCT for technical roles. Top UK law firms (Linklaters, A&O, CC, S&M, Hogan Lovells) use Watson-Glaser. Network Rail, Lloyd's of London, and Centrica use Pearson tests across their pipelines.

Scoring

Raven's tests report sten scores or percentiles relative to age-matched and education-matched norm groups. Watson-Glaser reports a raw score (out of 40 or 80 depending on version) plus percentile. BMCT reports a raw score and percentile against the comparison group (technicians, engineers, etc.).

Cutoff thresholds for trainee solicitor roles at top firms typically sit at 75th-85th percentile on Watson-Glaser. Civil Service Fast Stream typically requires top-quartile performance.

Preparation strategy

Raven's APM: Practice the seven core transformation rules until pattern identification is automatic. The classic APM is well-documented because of its age — buy a Raven's prep book, work through 100+ matrices, and you'll see the rule patterns repeat.

Watson-Glaser: The five sections require subtly different mindsets. Inference questions reward conservatism — "Probably True" beats "True" if there's any uncertainty. Deduction questions are more strict — only conclusions that must follow are correct. Practice with section-specific drills.

BMCT: Review high-school physics: Newtonian mechanics, simple machines (lever advantage, pulley counts), basic circuits, and fluid behaviour (pressure, flow). Most candidates fail BMCT not because the physics is hard but because it's been ten years since they touched it.

A worked BMCT-style question

Original and illustrative — not a real Bennett Mechanical item, but representative of the style:

BMCT questions like this reward genuine understanding of the underlying mechanical principle over memorised formulas, since the same gear-ratio logic reappears in pulley, lever, and belt-drive questions with different surface details each time — recognising the underlying pattern across different presentations is the actual skill being tested, not any single formula in isolation.

Which TalentLens test will you actually face?

Because Pearson TalentLens is a portfolio of several genuinely distinct instruments rather than one product, the single most useful thing you can do before preparing is confirm exactly which test your invitation refers to, since preparation for each is quite different. If your invitation mentions matrices, patterns, or "abstract reasoning," you're facing Raven's APM or APM-III or Adaptive Matrices — check whether it's timed (classic APM, 40 minutes) or shorter/adaptive (APM-III or Adaptive Matrices, 15-25 minutes), since the classic version rewards working through all items methodically while the adaptive versions reward getting early items right to unlock a fairer subsequent item set. If your invitation mentions critical thinking, reasoning, or is for a legal/policy role, you're almost certainly facing Watson-Glaser — see our dedicated Watson-Glaser guide for the full five-section breakdown with worked examples. If your invitation mentions mechanical, technical, or engineering aptitude, you're facing BMCT — the physics-refresher approach above is the right preparation path. Employers occasionally combine two of these in a single battery (most commonly Watson-Glaser paired with a Raven's variant for senior legal and policy roles), so don't assume a single-test invitation rules out a second stage using a different TalentLens product.

Why these tests are unusually well-documented

Because Raven's Progressive Matrices (1936) and Watson-Glaser (1925) are both close to a century old, they're among the most extensively studied instruments in the entire history of psychometrics — decades of published academic validity research exist for both, which is genuinely unusual among commercially deployed employment tests, most of which are newer and proprietary with far less public research behind them. This has a practical upside for candidates: because the underlying rule-sets and question formats have been stable for so long, third-party prep material for Raven's and Watson-Glaser tends to be more accurate and more thoroughly tested than prep material for newer, frequently-revised proprietary assessments, where publishers sometimes update the item bank faster than prep providers can track. BMCT, first published in 1940 and still updated periodically (BMCT-II is the current revision), sits in a similar position — mechanical-comprehension physics fundamentals don't meaningfully change over time even as the specific item pool is refreshed.

How TalentLens scores get used differently across sectors

Public-sector and security-adjacent employers (UK Civil Service, MI5/MI6/GCHQ) tend to use Raven's and Watson-Glaser scores as one input among several in a broader, more holistic selection process, rather than a single hard cutoff that eliminates candidates automatically. Private law firms, by contrast, often do apply a firmer Watson-Glaser percentile cutoff earlier in their process specifically because application volumes are extremely high relative to available training-contract places, making an early, defensible cognitive filter operationally necessary. Engineering and technical employers using BMCT similarly tend to treat it as a genuine pass/fail gate for safety-critical or technically demanding roles, since a candidate who struggles with basic mechanical-principle questions is a meaningfully different risk profile in those specific roles than in a general graduate scheme. Understanding which pattern your target employer likely follows — a holistic input versus a harder gate — is worth factoring into how much weight you put on maximising this specific score relative to the rest of your application.