TalentLens

TalentLens, Pearson's workplace psychometrics brand, publishes the Watson-Glaser critical-thinking test, Raven's Adaptive, BMCT-II and DAT Next Generation, which candidates take online and which are scored automatically against norm groups to produce percentile-based reports for hiring and development.
AssessmentScreening

HireAIScore rates TalentLens 46 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. TalentLens ranks 41 of 82 vendors rated in Assessment AI, against a category average of 46. That score is drawn from 29 evidence items across 7 rubric criteria for TalentLens — 24 with a cited source, 5 recording that nothing was located.

Score breakdown

TalentLens, criterion by criterion

Each rubric criterion is scored out of 100 on the cited evidence, then weighted into the total.

Weighted contribution shown to the right of each bar.

  1. Article 11 Technical Documentation

    Weight 20% · avg 50

    52

    +10.4

  2. Bias Audit Transparency

    Weight 18% · avg 41

    45

    +8.1

  3. FRIA Support

    Weight 15% · avg 31

    30

    +4.5

  4. Data Governance Disclosure

    Weight 15% · avg 53

    50

    +7.5

  5. Human Oversight Design

    Weight 12% · avg 55

    57

    +6.8

  6. Post-Market Monitoring

    Weight 12% · avg 38

    32

    +3.8

  7. Customer Documentation

    Weight 8% · avg 58

    58

    +4.6

Total weighted score

45.8(rounded to 46 for display)

Category avg is the mean raw score on that criterion across the 82 Assessment AI vendors in scope of this rubric, this one included.

Strongest criterion

Customer Documentation

58raw / 100

Contributes 4.6 to the total. 58 against a 58 category average.

Weakest criterion

FRIA Support

30raw / 100

Contributes 4.5 to the total. 30 against a 31 category average.

Cited evidence

The evidence behind the score

Every score is backed by at least one cited piece of evidence. Open a criterion to see what was found, when, and where.

Download diligence record

Evidence ledger

Items
29
Documentation
18
Audit report
1
Public statement
4
Product feature
1
Absence
5
With a source URL
24 of 29
Source hostnames
3
Fewest items
3
FRIA Support

These figures measure how thoroughly TalentLens was reviewed, not how TalentLens performed — an absence row, recording that nothing was located, is counted like any other item.

Article 11 Technical Documentation

5 items52
  • DocumentationCaptured Sep 25, 2026

    Pearson's April 2020 Watson-Glaser III Assessment Efficacy Report documents the test's item-bank design, five subtests and 18 norm groups, criterion validity (0.16 to 0.58 with job-performance indicators) and reliability (alpha 0.83, test-retest 0.73 to 0.89), and states that assessment reports of this kind are not independently assured.

    https://www.talentlens.com/content/dam/school/global/Global-Talentlens/uk/CriticalThinking/CTHub/Pearson_Efficacy_Watson-Glaser_2020.pdf

  • DocumentationCaptured Sep 25, 2026

    Pearson's Watson-Glaser efficacy page summarises the report's validity and reliability highlights and names the underlying samples (a business-school study, 123 legal-training students and 147 US working adults for internal consistency).

    https://www.pearson.com/en-gb/efficacy/product-evidence/watson-glaser.html

  • DocumentationCaptured Sep 25, 2026

    The Raven's Adaptive page explains that the computer-adaptive test starts at average difficulty, adjusts item difficulty after each answer and converts the resulting theta score into percentiles and other norm-referenced scores.

    https://www.talentlens.com/raven-adaptive.html

  • Product featureCaptured Sep 25, 2026

    The remote-proctoring page describes the HirePro integration's automated audio and video monitoring, face-mismatch detection against a verified photo and automatic logouts for browser infractions, but gives no technical detail on the models, error rates or data retention.

    https://www.talentlens.com/services/remote-proctoring.html

  • AbsenceCaptured Sep 25, 2026

    No current technical manual for any TalentLens test is posted on talentlens.com (173-URL sitemap, product pages, catalogues, guides), the legacy us.talentlens.com links to the BMCT-II manual and Watson-Glaser III FAQ redirect to the homepage, and no model documentation for the proctoring AI or ISO/IEC 42001 certificate was found.

    Have a URL that disputes this?

Bias Audit Transparency

4 items45
  • DocumentationCaptured Sep 25, 2026

    The efficacy report's fairness section gives Cohen's d ranges of -0.18 to 0.09 (disability), -0.06 to 0.37 (sex), -0.37 to 0.32 (ages 16-24 vs 45+), 0.28 to 0.75 (first language) and 0.75 to 1.31 (ethnic groups), and reports a differential-validity analysis in which scores predicted law-course performance consistently across ethnic groups.

    https://www.talentlens.com/content/dam/school/global/Global-Talentlens/uk/CriticalThinking/CTHub/Pearson_Efficacy_Watson-Glaser_2020.pdf

  • Audit reportCaptured Sep 25, 2026

    An independent BMJ Open study (Pinder et al., March 2023) of 629 applicants to 2021 UK public-health specialty training found that Black and Asian candidates were about 80% and 60% less likely than white British candidates to pass a psychometric stage made up of Watson-Glaser II, a numerical reasoning test and a situational judgement test, with similar differential attainment on each test.

    https://pmc.ncbi.nlm.nih.gov/articles/PMC10008157/

  • Public statementCaptured Sep 25, 2026

    TalentLens's hiring page is headed 'Make Bias Free Decisions' and says its psychometric assessments 'ensure unbiased decisions', a claim its own published ethnic score gaps do not support.

    https://www.talentlens.com/solutions/hiring.html

  • AbsenceCaptured Sep 25, 2026

    No NYC LL 144 bias-audit summary or third-party audit (Warden AI, Holistic AI, BABL AI, DCI, ORCAA) was found for any TalentLens test on talentlens.com or Pearson's efficacy pages; trust.warden-ai.com/talentlens, /pearson-talentlens and /pearson return 404, and trust.talentlens.com and talentlens.safebase.io do not resolve.

    Have a URL that disputes this?

FRIA Support

3 items30

Data Governance Disclosure

4 items50
  • DocumentationCaptured Sep 25, 2026

    The Security & Privacy page (last updated 13 Sep 2024) says Pearson's security policies are based on ISO 27001/27701 and NIST and names Talogy as subprocessor for the white-labelled TalentLens Online platform hosted in Ireland (Azure) and Indianapolis with no data transfer between instances, but still cites Talogy's EU-US Privacy Shield certification and leaves its privacy-statement and licence links as 'XX' placeholders.

    https://www.talentlens.com/about/security-and-privacy.html

  • DocumentationCaptured Sep 25, 2026

    The TalentLens privacy statement lists candidate data (identity, cognitive and personality data, work and education history), says de-identified aggregate answers may be used for research and product improvement, and keeps candidate data for as long as the customer holds an account or until the customer requests deletion.

    https://www.talentlens.com/about/security-and-privacy-privacy-statement.html

  • DocumentationCaptured Sep 25, 2026

    A French-language TalentLens Online FAQ says data is hosted in Dublin on Azure through Pearson's subprocessor PSI PAN, that some tests collect date of birth, sex and first language, that data is encrypted at rest (FIPS 140-2) and in transit (TLS), and that audit logs are kept for two years, while pen-test summaries are shared only under NDA and SOC 2 results not at all.

    https://www.talentlens.com/content/dam/school/global/Global-Talentlens/fr/autre/FAQ_TALENTLENSONLINE.pdf

  • Public statementCaptured Sep 25, 2026

    The Our Science page claims TalentLens holds a valid ISO 27001 certificate and Pearson holds Cyber Essentials Plus, and says the test platform is hosted by PSI Inc., but links no certificate or scope statement.

    https://www.talentlens.com/our-science.html

Human Oversight Design

4 items57

Post-Market Monitoring

4 items32

Customer Documentation

5 items58

Editorial notes

Company overview

TalentLens is Pearson's workplace psychometrics brand. Customers contract with Pearson Education Limited (Companies House 00872828, 80 Strand, London), and TalentLens sits in Pearson plc's Enterprise Learning and Skills division. It publishes long-established cognitive tests: the Watson-Glaser critical-thinking appraisal, Raven's Adaptive, the Bennett Mechanical Comprehension Test (BMCT-II), DAT Next Generation and NDIT. It also sells personality and motivation questionnaires, BPS test-user training and consulting such as validity studies and bespoke norms. Tests run on TalentLens Online, a white-labelled version of Talogy's platform with instances in Ireland (Azure) and Indianapolis. Scoring is automatic and norm-referenced: number-correct or adaptive theta scores become percentiles, T-scores, stanines and STENs against a standard or bespoke norm group. Watson-Glaser is sold for unproctored online screening and is widely used as an early sift in UK graduate and law-firm recruitment. The TalentLens tests themselves do not use machine learning. AI enters through optional HirePro remote proctoring, which analyses audio, video and face matches to flag possible cheating for the employer to review. TalentLens also resells Pearson's AI-scored Versant English test, which HireAIScore assesses separately as Versant by Pearson.

Regulatory exposure

TalentLens takes no public position on any hiring-AI law: none of the 173 pages in its sitemap mentions the EU AI Act, NYC Local Law 144, Illinois or Colorado. Its core scoring is psychometric norming, not machine learning. Answers are marked against fixed keys, item difficulties are calibrated in advance, and scores are turned into percentiles through norm tables; Raven's Adaptive and DAT Next Generation add adaptive ability (theta) estimation. Under the EU AI Act that most likely keeps the tests outside the Article 3(1) 'AI system' definition. Recital 12 and the Commission's February 2025 guidelines exclude systems that run on rules set solely by people, and simple statistical estimation. The optional HirePro proctoring, with its automated audio, video and face-match analysis, is plainly AI. It is the component exposed to Annex III point 4, with obligations now due on 2 December 2027. NYC is less clear-cut. LL 144 covers 'statistical modeling' that issues a score, and sifting NYC candidates on a Watson-Glaser pass mark alone meets DCWP's 'substantially assist' test. But DCWP's rules also require that a computer help identify the inputs or their weights, which fixed keys and norm tables arguably do not. TalentLens gives employers no audit or guidance to settle the question, and DCWP has promised tougher enforcement. California's FEHA rules, in force since 1 October 2025, expressly cover computer-based tests of skills and aptitude, so there is no machine-learning carve-out there. They also require four-year record retention, while TalentLens ties candidate-data retention to the customer's account. Illinois HB 3773 uses an inference-based AI definition close to the EU's, and Colorado's SB 26-189 disclosures start on 1 January 2027. In the EU, rejecting candidates automatically on a pass mark with no human review is also a solely automated decision under GDPR Article 22.

Path to a higher score

TalentLens already publishes more subgroup data than most test publishers, but nearly all of it covers one test and dates from 2020. The largest gains would come from three steps. First, publish the current technical manuals for Watson-Glaser III, Raven's Adaptive, BMCT-II and DAT Next Generation, or at least their fairness chapters, with subgroup results and impact ratios at typical pass marks. Second, commission an independent LL 144-style bias audit of Watson-Glaser with sex, race/ethnicity and intersectional impact ratios, and publish and repeat it annually. Third, state plainly how each regime classifies the tests (probably not an AI system under the EU AI Act, but an automated-decision system under California's FEHA rules), backed by deployer guidance on candidate notice, alternative assessments, pass-mark setting and record retention. The HirePro proctoring needs its own documentation: what the AI analyses, how biometric data is handled and for how long, error rates, and how candidates can challenge a flag. The privacy documents need reconciling: one named platform operator, current transfer mechanisms in place of the defunct Privacy Shield, a fixed retention period and working links instead of 'XX' placeholders. Operationally, a security.txt would bring talentlens.com under Pearson's disclosure policy, and a TalentLens component on status.pearson.com would give customers an incident history. Dropping 'bias free' claims that its own data contradicts would also help.

Regulatory frame

What applies to assessment AI.

Assessment is the most-regulated surface in AI hiring. NYC LL 144, EEOC and OFCCP frameworks, Illinois HB 3773, and EU AI Act Annex III §4 all apply, with overlap on bias-audit obligations.

Keep reading

All assessment vendorsHow this score is builtAnnex IV section-by-section: what AI hiring vendors must documentWhat makes a defensible NYC LL 144 bias audit

Compare

TalentLens against its nearest-scoring peers.

The closest scores in Assessment AI, side by side.

Build any comparison

Ranking

Others rated in Assessment AI

Ranked 41 of 82 by weighted total under rubric v1.0. The ordering is arithmetic on the rubric and carries no view on which tool suits a given hiring process.

  1. #38365Talents46F
  2. #39HackerEarth46F
  3. #40Korn Ferry46F
  4. #41TalentLensThis profile46F
  5. #42Talkpush46F
  6. #43Vervoe46F
All assessment vendors

Conflicts of interest

No vendor pays for placement, scoring, or removal. Casework - the consulting firm that operates this directory - provides paid services to some vendors. Any active or recent (within 24 months) commercial relationship is disclosed on the affected vendor profile and the review is reassigned to an independent reviewer. See the full policy on About.

Casework has no commercial relationship with this vendor.