No organisation sets out to reject job candidates because of their race, sex, age or sexual orientation. Yet applicants still have very different experiences depending on which store they apply to, which coordinator reads their CV, or whether they know someone on the inside. The examples of unfair hiring practices in this guide are the patterns that cause that – most of them invisible to the people running the process.
When you’re processing thousands of applications, small inconsistencies compound. They shrink your access to good candidates, damage your employer brand and, in some cases, expose you to legal risk.
This guide covers 14 common unfair hiring practices, explains why each one matters, shows where unfair tips into unlawful, and gives practical fixes you can start this month.
Hiring discrimination rarely comes from malice. It comes from inconsistency.
If different stores use different application forms, or one hiring manager runs a structured interview while another relies on “chemistry”, the system produces different outcomes for similar people. Nobody intended it, but the result is the same.
Proxy-driven decisions make it worse. When teams screen for brand-name employers, prestigious universities or unbroken employment, they crowd out evidence of job-related skills and disproportionately exclude career-switchers, carers and under-represented candidates. These shortcuts feel efficient because they play to the inherent biases everyone carries: we read familiar signals as quality.
Then there are communication black holes. When applicants hear nothing for weeks, the ones without insider confidence assume they’ve been rejected and move on.
The fix is process design: interview-first access, blind rubric-based scoring, predictable reminders and feedback for everyone, with audit trails that let you spot patterns before they become complaints.
An unfair hiring practice is any part of your process that gives some candidates a better or worse chance for reasons that have nothing to do with the job. Most unfair practices fall into five groups:
Not every unfair practice is illegal. A slow, disorganised process is unfair to everyone, but it isn’t discrimination. The line is crossed when a practice disadvantages people because of a protected characteristic – and that can happen without anyone intending it.
(This section summarises the UK position for orientation only. It isn’t legal advice. )
In Great Britain, the Equality Act 2010 protects job applicants as well as employees. It lists nine protected characteristics: age, disability, gender reassignment, marriage and civil partnership, pregnancy and maternity, race (including colour, nationality and ethnic or national origin), religion or belief, sex, and sexual orientation. Northern Ireland has separate legislation.
The Act recognises several types of discrimination. The two most relevant to hiring are:
| Direct discrimination | Indirect discrimination | |
| What it is | Treating someone less favourably because of a protected characteristic | Applying a rule to everyone that puts people who share a particular characteristic at a disadvantage |
| Hiring example | Rejecting an applicant because they mentioned a civil partnership, or because a manager wants someone “younger” | Requiring 10 years’ experience for a role that doesn’t need it, which screens out younger applicants; or “must be available any shift” with no flexibility, which disproportionately affects women with caring responsibilities |
| Can it be justified? | Generally no (age is the exception, if proportionate; occupational requirements are narrow) | Yes, if the employer can show it’s a proportionate means of achieving a legitimate aim |
| What to do | Remove the trigger: structured questions, blind first-pass scoring, recorded rationales | Test every requirement against the job’s real demands and drop the ones you can’t justify |
Two more duties matter in recruitment:
A candidate who believes they were treated unfairly because of a protected characteristic can bring a direct discrimination claim to an employment tribunal. They don’t need to have worked for you, and compensation for discrimination isn’t capped. Strict time limits apply and early conciliation through Acas usually comes first.
Australia and the US, briefly. In Australia, the Fair Work Act’s general protections cover prospective employees (refusing to employ someone can be “adverse action”), alongside federal laws on sex, race, disability and age discrimination and state and territory laws. In the US, Title VII, the Age Discrimination in Employment Act and the Americans with Disabilities Act are the main federal protections, enforced by the EEOC, with many states and cities adding their own rules (including salary history bans).
For a deeper look at the legal side, read our guide to discrimination in hiring and how to prevent it.
Once you can see unfair practices and hidden biases in your recruitment process, you can fix them. Here are 14 patterns we see often, and how to address each one.
How it shows up: Insider jargon, gender-coded words (“dominant”, “rockstar”), “must-haves” that are really nice-to-haves, no accommodation note, and wording that quietly signals a preference for a certain age – “recent graduate”, “digital native”, “young and energetic team”.
Why it’s unfair: These adverts put prospective candidates off before they apply, and not evenly. Carers, career-switchers, older applicants and disabled candidates are most likely to rule themselves out. Asking for a recent graduate when you mean “entry-level” can also amount to indirect age discrimination in the UK.
Fix: Rewrite job descriptions around outcomes. State the role’s primary functions, the reporting structure and the working pattern, then split essential requirements from optional ones. Use gender neutral language, add an accessibility statement with a named contact for adjustments, and translate for priority languages. If you mean “no experience needed”, say so.
How it shows up: Coordinators skim CVs to cut volume before anyone has assessed job-related skills, so hiring managers only ever see a filtered slice of the applicant pool.
Why it’s unfair: Pedigree proxies – brand-name employers, prestigious universities, polished formatting – beat actual capability. Without consistent criteria, two reviewers can rate the same CV very differently.
Fix: Start with a short, structured interview for every applicant, and hold CVs until blind first-pass scoring against a rubric is complete. For entry-level roles, consider removing CVs altogether. Our guide to blind hiring explains how to set this up. An interview-first platform like Sapia.ai runs as an overlay to your ATS, so you don’t have to replace your tech stack to do it.
How it shows up: Rejection notes like “didn’t feel right for us”, with no definition of what “right” means.
Why it’s unfair: “Culture fit” without a definition is where personal preferences influence hiring decisions most. In practice it often means “similar to me”, which narrows teams and can’t be audited.
Fix: Replace fit with values alignment you can measure. Describe your everyday work culture honestly – pace, autonomy, how disagreements get resolved – then turn it into four to six competencies. Write a scenario question for each, score answers against anchored rubrics as objective criteria, and add a second reader for borderline cases. Publishing those criteria internally is the start of a transparent hiring culture: anyone can see what “good” looks like and why a decision was made.
How it shows up: Each interviewer asks different questions, nobody uses a scorecard, and decisions come down to “chemistry” or “gut feel”.
Why it’s unfair: Different questions produce incomparable evidence, which makes subjective outcomes – and hiring discrimination – more likely.
Fix: Move to a structured interview process: the same questions per role family, behavioural rubrics, and scores and rationales captured for every candidate. Run calibration sessions so interviewers agree on what each score means. Structure helps both the interviewer and the candidate – the interviewer knows what to listen for, and the candidate knows they’re being judged on the same terms as everyone else. Read more on structured vs unstructured interviews.
How it shows up: Multi-hour tasks with vague success criteria and no accessible alternative.
Why it’s unfair: Long, unpaid tasks penalise shift workers, carers and disabled candidates, so drop-off rises sharply for those groups.
Fix: Keep work samples and job simulations to 30 minutes or less, tied directly to tasks the role involves. Publish instructions and marking criteria in advance, offer mobile-friendly and accessible versions, and score every submission against the same rubric.
How it shows up: Weekend availability, shift flexibility or minimum hours are applied differently from one location to the next.
Why it’s unfair: Candidates are treated differently for the same role depending on where they apply. Because the rules aren’t written down, it’s almost impossible to audit.
Fix: Build a central policy library. Ask eligibility questions the same way for everyone, at the same stage, and log any exception with a clear rationale.
How it shows up: Referral schemes or a manager’s personal network create side doors straight to a manager screen or an offer.
Why it’s unfair: It prioritises candidates based on who they know, not what they can do. If your current workforce is homogeneous, referrals tend to reproduce it.
Fix: Keep referrals – they’re valuable – but route every candidate through the same structured, blind first assessment. Compare referral and non-referral outcomes monthly, and tell other employees how the scheme works so it doesn’t look like favouritism.
How it shows up: Asking what candidates earn now, then basing the offer on it.
Why it’s unfair: Pay anchoring carries historical pay gaps into new offers. In many US states and cities, asking about salary history is banned outright.
Fix: Publish salary bands and base offers on the band and job-related factors only. Remove salary history fields from your forms and train hiring managers to make consistent offers.
How it shows up: Desktop-only forms, mandatory high-bandwidth video, timed tests and no screen-reader support.
Why it’s unfair: These barriers can amount to disability discrimination, and they also exclude people for reasons like not owning a laptop or having a weak connection.
Fix: Build a mobile-first, text-friendly process that works with screen readers and in low-bandwidth conditions. Publish a clear route for requesting reasonable adjustments, name who handles requests, and respond quickly.
How it shows up: No acknowledgement after applying, no timeline, and candidates chasing for updates.
Why it’s unfair: Silence drives disproportionate drop-off among candidates without insider confidence, especially in competitive markets where they have other options.
Fix: Set expectations in your first message, send expiry-aware reminders by SMS and email, and give every candidate feedback at decision time.
How it shows up: Candidates below a score threshold are auto-rejected, with no rubric or explanation stored.
Why it’s unfair: Neither candidates nor auditors can see why a decision was made. Candidates may reasonably assume they were rejected because of a disability, their age or their race, and you have nothing to show otherwise.
Fix: Base any automation on structured questions and anchored rubrics, store the rationale for every decision, and keep a human in the loop for threshold cases.
How it shows up: Personal details in free-text notes, uncontrolled downloads and unclear retention schedules.
Why it’s unfair: Weak data governance exposes vulnerable groups most and erodes trust in your process.
Fix: Redact identifiers for first-pass review, use role-based access and regional data residency, enforce retention schedules and keep exportable audit logs. These steps build trust and help you meet data protection requirements.
How it shows up: A vacancy is posted as a formality when the manager has already chosen someone, criteria are written around one person’s CV, or promotions happen through a quiet tap on the shoulder.
Why it’s unfair: Unfair internal hiring practices waste the time of everyone who applies and tell other employees that progression depends on proximity to the right manager. If informal networks are homogeneous, they can also produce indirect discrimination over time.
Fix: Advertise internal roles to every eligible employee for a fixed window and use the same structured assessment for internal and external applicants. If you decide to make a direct appointment, say so openly instead of running a fake process, and record why.
How it shows up: Knock-out questions such as “currently employed?”, automatic rejection of gaps longer than six months, or recruiters quietly favouring people who are already in work.
Why it’s unfair: Filtering on current employment status removes potentially qualified applicants for reasons that say nothing about their ability – redundancy, caring, illness, study or relocation. Gaps are more common among women and disabled people, so these filters can also create indirect discrimination risk.
Fix: Remove employment status and gap filters from your screening rules. If recency genuinely matters – a licence or certification that lapses, for example – ask about that requirement directly.
Most discriminatory interview questions come from managers making small talk or trying to judge reliability. The safe rule: every question should test the applicant’s suitability for the job, and nothing else. Here are the questions to steer clear of, and what to ask instead.
| Question to avoid | Why it’s risky | Ask instead |
| “Are you married?” / “Do you have children or plan to?” | Marriage and civil partnership, pregnancy and maternity, sex | “The role includes two weekend shifts a month. Can you work the rota as described?” |
| “How old are you?” / “When did you leave school?” | Age – and it signals a preference for a certain age | “Do you hold the [licence/certification] the role requires?” |
| “Where are you from originally?” / “Is English your first language?” | Race, including national origin | “The role involves writing incident reports. Tell us about a time you wrote one.” |
| “Do you have any health conditions or disabilities?” | Disability – in the UK, pre-offer health questions are restricted except in limited cases, such as arranging adjustments for the interview | “Is there anything we can adjust in the process to help you do your best?” |
| “Is that your legal name?” / questions about transition | Gender reassignment | Run right-to-work checks the same way, at the same stage, for every candidate |
| “Will you need time off for religious holidays?” | Religion or belief | “Can you work the published rota, including [days]?” |
| “Are you working at the moment?” / “Why the gap?” | A proxy that disadvantages carers, disabled people and those made redundant | “Tell us about a time you [competency the role needs].” |
| “What do you earn now?” | Pay anchoring; banned in parts of the US | “The band for this role is [£X–£Y]. Does that work for you?” |
The simplest way to stop these questions is to give interviewers a fixed question set and a scorecard. If the question isn’t on the card, it isn’t asked.
Fixing unfair practices is easier to justify when you can see the result. Three Sapia.ai customers show what happens when the first step is the same structured, blind interview for everyone. (Figures are as published in each customer’s case study.)
LNER (UK rail). LNER’s challenge included “underrepresentation of diverse candidates”. It removed CV uploads and video interviews from the application journey and gave hiring managers pre-vetted shortlists ranked on role fit. The result was “a consistent 30% ethnic minority representation through all hiring stages” – evidence that racial diversity in the applicant pool survived shortlisting, rather than thinning out at each step. LNER also reported a 97% completion rate, 9/10 candidate satisfaction and a 57% decrease in time to hire. Read the LNER case study.
Costa Coffee (UK retail). Facing 300,000 applications for 8,000 openings, Costa found that “manual, inconsistent decisions let unconscious bias creep in between stores” – example 6 above, at scale. Its Chat Interview is “untimed, mobile-first, and blind to gender, age, ethnicity, or disability”, and works with screen readers and voice-to-text tools. Costa reports an 89% Chat Interview completion rate and 8.8/10 candidate satisfaction from 180,000 interviews. Read the Costa Coffee case study.
Qantas Graduate Program (Australia). Qantas received a record 4,500+ graduate applications, and found that “for graduates, CVs rarely reveal meaningful differences”. Instead, graduates answer five open-ended, competency-based questions on mobile, in their own time, and every applicant receives personalised feedback. Qantas reports a 96.7% completion rate, 8.8/10 candidate satisfaction and 1,831 recruiter hours saved in the 2025 intake. Read the Qantas case study.
None of these show bias “removed”. What they show is that a consistent, structured first step can keep representation steady, keep completion high and give every candidate the same chance.
You don’t need months to find unfairness in your process. These three checks will give you a better understanding of where to start, in 30 minutes each.
Pull the last 90 days of recruitment data and look for sharp drops between application and interview completion, and between interview and offer. Break it down by site or brand. If one location shows a steep decline that others don’t, investigate – it usually signals different standards or different communication.
Where you hold voluntary diversity monitoring data, compare pass rates between groups. Illustrative example: 400 applicants from group A reach interview and 120 pass (30%). 200 from group B reach interview and 40 pass (20%). Group B’s rate is 20 ÷ 30 = 67% of group A’s – below the 80% “four-fifths” threshold, so that stage is worth a closer look. A ratio below 80% isn’t proof of discrimination, but it tells you where to dig.
Compare average rubric scores across reviewers and hiring managers. A wide spread usually means a calibration problem: people are interpreting your criteria differently. If one interviewer scores much higher or lower than the rest, run a calibration session using real (anonymised) answers.
Find the candidates who’ve waited longest at each stage. Long silences drive disengagement and perceived unfairness, and candidates who hear nothing for weeks usually assume they’ve been rejected and accept other offers.
Track these metrics to measure progress towards a fair, inclusive hiring process:
Here’s how to pilot changes, prove they work and prepare to scale in 30 days:
Use this checklist to track progress as you roll out changes:
Most unfair hiring is a thousand paper cuts, not one villain. Sapia.ai addresses the cuts at the front of the funnel.
What Sapia.ai does: Its Chat Interview gives every applicant the same structured, mobile-first, untimed interview at the point of application, before anyone sees a CV. Answers are scored against consistent criteria, hiring managers get an explainable shortlist, and every candidate receives personalised feedback. It runs as an overlay to your existing ATS, and diversity dashboards let you monitor representation by stage. That covers examples 2, 4, 5, 9, 10 and 11 directly.
What it doesn’t do: Sapia.ai won’t rewrite your job adverts, fix a referral policy, stop managers making informal internal appointments or replace legal advice. It doesn’t replace your wider diversity, unconscious bias and inclusion programmes, and it doesn’t make the final hiring decision – your managers do. Building an inclusive workplace still depends on what happens after the shortlist.
If you want to see how an interview-first process would work on your highest-volume role, book a demo of Sapia.ai.
Common examples include screening CVs before assessing skills, unstructured interviews where each candidate gets different questions, rejecting people for poor “culture fit” without defining it, letting referrals or internal favourites skip steps, asking about salary history, and filtering out applicants with employment gaps. Each one gives some candidates a better or worse chance for reasons unrelated to the job.
Not always. A slow or disorganised process is unfair but not necessarily unlawful. It becomes discrimination when a practice treats someone less favourably because of a protected characteristic (direct discrimination), or when a rule applied to everyone disadvantages a group that shares a protected characteristic and can’t be justified (indirect discrimination).
Direct discrimination is treating a candidate less favourably because of a protected characteristic, such as rejecting someone because of their age or religion. Indirect discrimination is applying a neutral-looking requirement that puts a group at a disadvantage, such as an unnecessary experience requirement that screens out younger applicants. Under the Equality Act, indirect discrimination can be justified if it’s a proportionate means of achieving a legitimate aim; direct discrimination usually can’t.
Avoid questions about marital status, children or pregnancy plans, age, national origin, religion, health or disability, and current salary. Instead, ask whether the candidate can meet the job’s specific requirements, such as a shift pattern or certification. A fixed question set with a scorecard is the most reliable way to keep interviews on job-related ground.
The four-fifths rule is a rule of thumb from US EEOC guidance for spotting adverse impact. If one group’s selection rate is less than 80% of the rate for the group with the highest selection rate, that stage may be having an adverse impact and should be investigated. UK law has no fixed threshold, but the calculation is still a useful early warning.
Generally, no. Choosing a candidate because of a protected characteristic, such as through a quota, is unlawful under the Equality Act. Positive action is allowed: employers can encourage applications from under-represented groups and, where two candidates are as qualified as each other, may prefer the one from an under-represented group in limited circumstances.