FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: HR consultants
AI can support redundancy scoring, but selection needs meaningful human involvement or Article 22C safeguards, and criteria must not discriminate. A flawed matrix can generate several claims at once, with the adviser who designed it in the firing line.
Part of: AI in HR consultants
In short
Analytics tools can rank employees for redundancy from performance, attendance and productivity data. Selection leads to dismissal, a significant decision, so if the algorithm effectively decides, Article 22C safeguards apply and the ICO says review must be active, not tokenistic. Data-driven criteria can penalise disabled, part-time or recently returned staff. Manjang v Uber Eats shows how automated processes lead to discrimination claims. One flawed matrix can produce several claims, and the client may pursue the adviser. PI usually responds, subject to the wording.
When a client needs to reduce headcount, HR consultants design the selection pool, the criteria and the scoring matrix. AI and analytics tools can offer to do much of this: pulling performance, attendance, skills and productivity data, weighting it and ranking employees. Some tools recommend who should be selected. For a client under pressure, a ranked list that appears objective is attractive.
Selecting someone for redundancy leads to dismissal, which is a significant decision under UK GDPR Article 22A. If selection is effectively made by the algorithm, with managers only approving the list, it is likely to be solely automated and the Article 22C safeguards apply: inform the employees, let them make representations, and allow human intervention and challenge. The ICO's draft ADM guidance says human review must be active, not tokenistic, and that reviewers must understand the system's limits. A manager who cannot explain why one employee scored lower than another has not meaningfully reviewed the decision.
Fair redundancy selection also requires objective criteria applied fairly, genuine consultation, and the chance for employees to challenge their scores. An opaque algorithm makes all three harder.
Consultation is where this often unravels. Employees and their representatives are entitled to understand how they were scored. If the employer can only point to a tool's output and cannot say how each score was reached, it is poorly placed to consult meaningfully, answer appeals or show a tribunal that the selection was fair.
Selection criteria built from data can discriminate indirectly. Attendance data can penalise disabled staff or those with caring responsibilities. Productivity data can penalise part-time workers or those on adjusted duties. Data from a period of maternity leave can distort scores.
The Manjang v Uber Eats case is a warning about automated processes affecting people's work. Facial recognition checks repeatedly failed to verify a Black courier and an automated process removed him from the platform. The tribunal refused Uber's application to strike out the claim, and the EHRC-funded case settled in 2024. The EHRC's message: be transparent about AI, explain decisions, provide a way to challenge them, and guard against discrimination.
The CIPD Code requires fair, consistent and equitable policies (positive and active impact, cl.1), equal opportunity (cl.2), judgement free from bias (good character, cl.5) and an evidence-based service (professional service and competence, cl.2).
Illustrative: a consultant builds a redundancy matrix using an analytics tool that weights attendance and output. Managers accept the ranking without review. Several selected employees had disability-related absence or were on reduced hours. They bring unfair dismissal and discrimination claims, arguing the scoring was opaque and they could not challenge it. The client settles and pursues the consultant who designed the matrix.
Redundancy exercises affect several people at once, so a flawed matrix can generate multiple claims from one piece of advice. PI usually responds to a client's claim that the matrix design was negligent, subject to the policy terms. Check how the policy aggregates related claims, whether discrimination-related losses are covered, and how your contract limits liability. The client's EPL policy may respond to the employees' claims first. See PI versus client EPL.
Apex arranges PI for HR consultants who run redundancy and restructuring work, and a broker can check how your policy handles multiple claims from one exercise. Related: AI in performance reviews and AI in HR consulting.
An algorithm can support selection, but if it effectively makes the decision with no meaningful human involvement, Article 22C safeguards apply. Fairness also requires objective criteria, consultation and a real chance for employees to challenge their scores.
It shows that automated processes affecting people's work can lead to discrimination claims. The EHRC urged transparency about AI, explanations for decisions, a way to challenge them, and safeguards against discrimination.
Absence linked to disability, pregnancy or caring can disadvantage protected groups. Using raw attendance data without adjustment can make selection indirectly discriminatory and harder to defend.
It can lead to several employee claims against the client, which may then become one or more recovery claims against the adviser. How the PI policy aggregates related claims affects the limit and excess.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
