Trust centre
We measure leadership and sales capability rigorously. We protect your data, your candidates, and your hiring process with equal rigour.
Data security and privacy
Every client organisation has its own isolated space on the platform. Candidates, responses, scores and reports belong to a single organisation, and access rules are enforced by the database itself rather than by the interface. One client cannot see another client's people, results or activity, by accident or by design.
All data is encrypted in transit using TLS 1.2 or above, and encrypted at rest. Assessment links are tokenised, so no candidate identifier ever appears in a URL, and we store only a hash of each token. Uploaded material such as interview transcripts is held in private storage that is unreachable without an authenticated, authorised session.
We process personal data in line with the UK GDPR, the EU GDPR and the Data Protection Act 2018, and in line with employment and equality law in the jurisdictions where our clients hire. Clients are the data controller for their candidates; Mint Psychometrics acts as the processor, working only on documented instructions.
Every access is written to an append-only audit trail. We do not sell data, we do not use candidate responses to train third-party models, and we do not share personal data with third parties beyond the infrastructure providers required to run the service. Full detail, including retention periods and how candidates exercise their rights, is set out in our Privacy Notice.
No route exists between one organisation's data and another's. Benchmarks are anonymised and aggregated, and small cohorts are suppressed so nobody can be identified.
Fair assessment practices
Structured assessment exists partly to reduce bias. Every candidate answers the same material, in the same order of difficulty, against the same published scoring rules. That consistency is what makes a decision explainable later, and it is the single most effective control we have against judgements that drift with the person in the room.
Assessment content is reviewed for cultural neutrality and plain language. We avoid idiom, regional references, sporting metaphors and anything that rewards familiarity with one national working culture over another. Content is written to be readable by a competent non-native English speaker, and reviewed again whenever the item bank changes.
Demographic information — gender, ethnicity, age, disability status and similar — is captured separately from assessment data, and is never an input to scoring, to our rules engine or to any AI model. It exists only so that organisations can monitor their own hiring for adverse impact, which is a legal duty in many jurisdictions and good practice everywhere.
Reasonable adjustments are supported as standard. Candidates can pause the timer and return later, extended time can be granted, and the assessment can be re-issued where a genuine technical or health issue interrupted a sitting. Adjustments are recorded against the sitting so a reviewer can interpret the result correctly.
No automated decisions
Our tools never accept or reject anyone. They give structured evidence to the people accountable for the decision.
Written responses are scored with AI assistance against a published rubric. That output is a draft and is treated as one. Nothing reaches a decision-maker until a trained reviewer has read the evidence and either confirmed the draft or overridden it. An override requires a written rationale, which is stored with the result.
Risk signals — integrity concerns, unusual response patterns, inconsistent evidence — are surfaced for human attention. They never trigger an automatic rejection, and they never adjust a score on their own. A flag is a prompt to look closer, not a verdict.
Every material action is logged: score overrides and their rationales, report approvals and exports, assessment re-issues, and each sign-in to the platform.
Audit and accountability
The platform keeps an append-only audit trail: who accessed a candidate record, when, from which account, and what they did. Records cannot be edited or deleted by users, including administrators. If a question is raised months later about how a decision was reached, the history is there to answer it.
Measurement frameworks are defined in advance, scoring rules are consistent across every candidate, and each version of the item bank and rubric is recorded against the sittings it scored. That means a result can always be reproduced against the rules that were live at the time, rather than the rules in force today.
Scoring rules are documented and can be shared with external auditors, occupational psychologists or legal advisers where a client needs independent assurance. Reasonable adjustments, re-sits and their justifications are recorded openly in the same trail, so an accommodation never looks like an anomaly.
Candidate support and accessibility
Candidates are told what to expect before they begin: how long the assessment takes, what each section asks of them, how their answers are used, and who sees the result. Guidance stays available while they work, and a public information page explains our approach in plain language without anyone needing to sign in.
Accessibility accommodations are available. Candidates can pause and return, extended time can be arranged, the interface works with keyboard navigation and screen readers, and colour is never the only way information is conveyed. If someone needs an adjustment we have not anticipated, we would rather be asked than have them struggle through.
A technical issue button is present on every assessment screen at all times, so a candidate can report a problem without leaving the assessment or hunting for an email address. Reports reach our team directly and are triaged the same day where possible. The experience is designed to be clear and fair — the difficulty should live in the thinking, never in the software.
Our commitment
Structured assessment does not replace human judgement. It gives human judgement something honest to argue with. The organisations that get the most from our work are the ones that treat a profile as evidence to interrogate, not an answer to accept.
We are conservative about what we claim and open about how we work. Our frameworks, scoring rules and governance practices are reviewed regularly, and we update them when the evidence says we should. Where we get something wrong, we would rather hear it early.
If you have questions about our security posture, our fairness approach, or how a specific result was produced, ask us directly. We will answer with detail rather than reassurance.