Versant by Pearson
Versant by Pearson is a suite of AI-scored English proficiency tests (plus spoken Spanish, French, Arabic and Dutch) that automatically score candidates' spoken and written responses on Pearson's Global Scale of English within minutes, used by employers such as contact centres and BPOs to screen applicants and benchmark staff through the ScoreKeeper dashboard.
HireAIScore rates Versant by Pearson 49 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Versant by Pearson ranks 23 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 Versant by Pearson — 23 with a cited source, 6 recording that nothing was located.
§ 01 - Company facts
- Legal name
- Pearson plc
- Founded
- 1844
- Headquarters
- United Kingdom · GB
- Pricing tier
- Mid
- Market side
- Employer-side (AEDT)
- Categories
- AssessmentScreening
- Website
- pearson.com
These are company details, not findings. They are editable by a verified representative and carry no weight in the score, which is built only from the cited evidence below.
§ 02 - Score breakdown
§ Score breakdown
Category scoring
Weighted contribution shown to the right of each bar.
- 01
Article 11 Technical Documentation
Weight 20%63
+12.6 · category avg 50
- 02
Bias Audit Transparency
Weight 18%42
+7.6 · category avg 41
- 03
FRIA Support
Weight 15%30
+4.5 · category avg 31
- 04
Data Governance Disclosure
Weight 15%53
+7.9 · category avg 53
- 05
Human Oversight Design
Weight 12%56
+6.7 · category avg 55
- 06
Post-Market Monitoring
Weight 12%43
+5.2 · category avg 38
- 07
Customer Documentation
Weight 8%61
+4.9 · category avg 58
Category avg is the mean raw score on that criterion across the 82 Assessment AI vendors in scope of this rubric, this one included.
§ 03 - Strongest · weakest
Strongest category
Article 11 Technical Documentation
Raw score 63 · contributes 12.6 to total.
63 against a 50 category average
Weakest category
FRIA Support
Raw score 30 · contributes 4.5 to total.
30 against a 31 category average
§ 04 - Cited evidence
Download diligence record→§ Evidence
Cited per category
Every score is backed by at least one cited piece of evidence.
Evidence ledger
- Items
- 29
- Documentation
- 20
- Public statement
- 3
- Absence
- 6
- With a source URL
- 23 of 29
- Source hostnames
- 5
- Fewest items
- 3
- FRIA Support
These figures measure how thoroughly Versant by Pearson was reviewed, not how Versant by Pearson performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
4 items63
- DocumentationCaptured Sep 25, 2026
Pearson's 2024 validation summary for the Versant English Speaking and Listening Test documents six task types, a speech recognizer optimised for learner accents, content and manner-of-speaking scoring, Rasch calibration on responses from nearly one million tests, GSE scaling by a ten-expert panel, and a 260-person held-out sample with a .96 Overall machine-human correlation.
- DocumentationCaptured Sep 25, 2026
The 2024 Professional English Test validation report explains the equal 25% weighting of four skills and how models trained on expert human ratings machine-score spoken and written responses (including latent-semantic-analysis content scoring), but it carries over machine-accuracy evidence from a 124-candidate PTE Professional study instead of re-running it.
- DocumentationCaptured Sep 25, 2026
Pearson's 2025 Strategic Report says its Responsible AI Principles and AI Governance Framework were set up in 2025 and that the framework aligns to the NIST AI RMF, the EU AI Act, ISO 42001/27001/27701 and SOC 2, which is a group-level alignment claim rather than a Versant-specific certification or technical file.
https://plc.pearson.com/sites/pearson-corp/files/annual-reports/2025/strategic-report-2025.pdf
- AbsenceCaptured Sep 25, 2026
No Versant-specific model card, instructions-for-use document, EU AI Act technical documentation pack or ISO/IEC 42001 certificate was found on the Versant HR and Our Tests pages, in the five validation reports and official test guides, in the Pearson support knowledge base, or in Pearson's privacy center and AI FAQ.
Bias Audit Transparency
5 items42
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144 bias-audit summary, impact-ratio table or independent algorithmic audit was found for any Versant test on Pearson's Versant HR and Our Tests pages, in the validation reports and test guides, or on the support site, and no Warden AI trust page exists for Pearson or Versant (both URLs return 404).
- DocumentationCaptured Sep 25, 2026
The Speaking and Listening validation report lists its sample by gender (123 male, 135 female, 2 non-binary) and 30 first languages and says external linguists reviewed items to be free of 'discrimination, prejudices or biases', but it reports no subgroup score comparison, adverse-impact or DIF analysis even as it claims conformance with the EEOC Uniform Guidelines.
- Public statementCaptured Sep 25, 2026
Pearson markets Versant as 'bias-free testing' whose AI assesses proficiency 'without human bias', but no published subgroup fairness analysis backs these claims.
https://www.pearson.com/languages/hr-professionals/versant/our-tests.html
- DocumentationCaptured Sep 25, 2026
An independent review of the Speaking and Listening Test by Iowa State University researchers led by Gary Ockey and Carol Chapelle (Language Assessment Quarterly 23(1), January 2026) judged four of its six tasks effective for the oral-communication construct, but it is a closed-access construct review, not a bias audit.
https://api.openalex.org/works/doi:10.1080/15434303.2025.2608742
- DocumentationCaptured Sep 25, 2026
Versant's sample score report says the Manner of Speaking sub-score depends on pronouncing English sounds close to typical norms 'without too much of an accent', which makes subgroup analysis by first language and national origin especially important.
FRIA Support
3 items30
- AbsenceCaptured Sep 25, 2026
No Fundamental Rights Impact Assessment template, EU AI Act Article 26/27 deployer guidance or AI Act statement for Versant was found on the Versant HR and Our Tests pages, in the validation reports and test guides, in the Pearson support knowledge base, or in Pearson's privacy center and AI FAQ.
- DocumentationCaptured Sep 25, 2026
The validation reports give partial deployer guidance: they endorse score use only when test-taker identity is verified, tell score users to set their own cut scores from localized research, and offer Pearson tools and assistance in setting cut scores.
- Public statementCaptured Sep 25, 2026
Pearson's AI FAQ says its teams 'conduct impact assessments' and maintain human-in-the-loop governance, but no impact assessment or deployer-facing summary is published for Versant.
https://www.pearson.com/privacy-center/resources/ai-frequently-asked-questions.html
Data Governance Disclosure
4 items53
- DocumentationCaptured Sep 25, 2026
The Professional English Test report describes scoring-development data inherited from PTE Professional field tests: 73 native speakers plus 1,695 writing and 973 speaking learners from 46 countries, more than 200,000 spoken and 50,000 written responses, and over 118,000 transcriptions by trained US-based transcribers, with ratings by raters holding master's degrees.
- DocumentationCaptured Sep 25, 2026
The Speaking and Listening report says none of the 260 validation participants' responses were used to train the speech-recognition or scoring models.
- DocumentationCaptured Sep 25, 2026
Pearson's group privacy notice (updated 17 August 2026) says spoken responses are recorded for scoring, certain audio, video and images are kept for up to six months after a test, AI-based voice and face matching is used for test security, and anonymised and pseudonymised data are used to train Pearson AI models.
https://www.pearson.com/privacy-center/privacy-notices/full-privacy-notice.html
- AbsenceCaptured Sep 25, 2026
No Versant-specific data-retention schedule, public DPA, subprocessor list or ISO 27001/SOC 2 attestation scoped to Versant or ScoreKeeper was found on the Versant pages, the ScoreKeeper site or Pearson's privacy center, and ScoreKeeper's privacy link only resolves to Pearson's generic privacy-notice index.
Human Oversight Design
5 items56
- Public statementCaptured Sep 25, 2026
Pearson's FAQ says Versant scores are '100% AI-based' and humans are involved only in training the scoring models, so by default no human reviews individual scores.
https://www.pearson.com/languages/hr-professionals/versant/our-tests.html
- DocumentationCaptured Sep 25, 2026
Pearson's ScoreKeeper account-types article (dated 3 February 2026) says Test Administrators can see results, listen to sample responses (if enabled) and decide whether to share score reports, and candidates get results only if the administrator chooses to share them.
https://support.pearson.com/languages/s/article/Versant-Account-types?language=en_US
- DocumentationCaptured Sep 25, 2026
ScoreKeeper's remote-monitoring feature combines AI flagging of suspicious activity with proctor and audio review that customers carry out after the test inside ScoreKeeper.
https://www.pearson.com/languages/hr-professionals/versant/our-tests.html
- DocumentationCaptured Sep 25, 2026
Pearson's privacy notice says its scoring uses automated decisions and gives individuals the right to contest them and require human review via dataprivacy@pearson.com, and it says human reviewers check proctoring flags before any action is taken.
https://www.pearson.com/privacy-center/privacy-notices/full-privacy-notice.html
- DocumentationCaptured Sep 25, 2026
The sample Speaking and Listening score report explains each GSE and CEFR result with plain-language level descriptors and improvement tips, which helps the employer interpret results rather than handing down a pass/fail verdict.
Post-Market Monitoring
3 items43
- DocumentationCaptured Sep 25, 2026
Pearson's status page lists a Versant component under English Language Learning, but the Versant-filtered incident history shows only five entries between December 2023 and October 2025, mostly Pearson-wide login and courseware outages plus one ScoreKeeper maintenance window, and none has a postmortem.
- DocumentationCaptured Sep 25, 2026
Pearson's Responsible Security Disclosure Policy offers a responsible.disclosure@pearson.com reporting channel but limits its scope to products whose root domain publishes a security.txt file, and none was found for pearson.com or versanttest.com.
https://www.pearson.com/legal-information/our-policies/responsible-security-disclosure-policy.html
- AbsenceCaptured Sep 25, 2026
No scoring-engine changelog, model-version history, or score-drift or fairness-monitoring dashboard was found for Versant; the validation reports are static 2024 documents, and Pearson publishes no post-deployment performance data by candidate group.
Customer Documentation
5 items61
- DocumentationCaptured Sep 25, 2026
Pearson's Versant 'Our tests' page describes five English tests and their intended users (corporations, call centres, BPO providers), links a test description and validation report for each, and has an FAQ covering automated scoring, reporting and integrations.
https://www.pearson.com/languages/hr-professionals/versant/our-tests.html
- DocumentationCaptured Sep 25, 2026
GSE Job Profiles gives employers expert-validated English benchmarks for nearly 1,400 job roles so they can set role-specific requirements and assess candidates with Versant.
https://www.pearson.com/languages/why-pearson/the-global-scale-of-english/job-profiles.html
- DocumentationCaptured Sep 25, 2026
Pearson's June 2024 GSE Job Profiles research summary reports industry panels that derived cut scores for roles, including customer service agents, from Versant test responses, and says the resulting ranges are guidelines for companies to set their own local benchmarks.
- DocumentationCaptured Sep 25, 2026
The candidate-facing official guide for the Speaking and Listening Test tells test-takers the test is 'autoscored', that all spoken answers are recorded for scoring, and that they will be told in advance about remote monitoring.
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144, Illinois HB 3773, Colorado or EU AI Act deployer guidance, candidate-notice template or public DPA for Versant was found on the Versant HR, Our Tests and test-taker pages, in the validation reports and guides, in the support knowledge base or in Pearson's privacy center, and the test-taker landing and preparation pages do not mention AI scoring.
§ 05 - Editorial notes
Company overview
Versant by Pearson is the workplace language-testing line of Pearson plc, the London-headquartered education company founded in 1844, and sits in Pearson's English Language Learning business alongside PTE. The technology came from Ordinate Corporation, which combined with Pearson in January 2008. Versant sells short, fully automated tests: the 17-minute English Speaking and Listening Test, a 4-Skills Essentials test, a 35-minute Writing Test, the Professional English Test and the English Placement Test, plus spoken tests in Spanish, French, Arabic and Dutch. Speech and text responses are machine-scored and reported within minutes on Pearson's 10–90 Global Scale of English (mapped to CEFR) in the ScoreKeeper admin dashboard, which also offers AI-flagged remote monitoring. Pearson says it delivers about 3 million tests a year to more than 1,200 businesses, governments and universities, and it markets Versant to corporations, call centres and BPO providers for high-volume hiring.
Regulatory exposure
When employers screen applicants by applying cut-offs to Versant scores, the tests are high-risk AI under EU AI Act Annex III point 4 (obligations now due 2 December 2027). They are also likely an AEDT under NYC Local Law 144, where DCWP has promised tougher enforcement after the December 2025 Comptroller audit. Illinois HB 3773 has required notice of AI use since 1 January 2026, but Versant's test-taker web pages do not mention AI scoring. Colorado's SB 26-189 adds disclosure duties from 1 January 2027. California's FEHA rules require employers to keep records for four years, while Pearson's privacy notice caps storage of certain test audio at six months. Fairness is the biggest exposure. Pearson calls Versant 'bias-free' and claims it conforms to the EEOC Uniform Guidelines, but it publishes no subgroup or adverse-impact data, even though its Manner of Speaking score rewards pronunciation close to 'typical norms'. In Pearson's favour, it says proctoring flags do not track emotional behaviour, and it offers human review of automated decisions.
Path to a higher score
Pearson's quickest gains are in showing fairness evidence. It should publish subgroup analyses for every Versant test: score gaps, impact ratios and DIF by sex, age, race/ethnicity where lawful, and first language. It should also commission an independent bias audit that NYC employers can rely on, repeat it every year, and stop calling the tests 'bias-free' until that evidence exists. Its strong validation reports could become EU-AI-Act-style instructions for use plus a Versant model card covering engine versions, training data and known limits (accents, speech impairments, audio quality), backed by ISO/IEC 42001 certification scoped to Versant. Deployer materials are the other gap: candidate-notice templates for NYC, Illinois and Colorado, FRIA and Article 26 guidance, workflows for human review and accommodations, and cut-score guidance that includes adverse-impact checks. The data-governance and monitoring scores would rise with a Versant-specific retention schedule and DPA, a clear answer on whether candidate audio trains Pearson models, a scoring-engine changelog, and a security.txt on versanttest.com.
§ 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.
§ Compare
Build any comparison→Versant by Pearson against its nearest-scoring peers.
In Assessment AI.
§ Others rated in Assessment AI
All assessment vendors→Ranked 23 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.
- 20SHL51F
- 21Xobin50F
- 22Test Partnership50F
- 23Versant by PearsonThis profile49F
- 24Jobma49F
- 25Infor Talent Science49F
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.