Picture two hours spent on a job application. Careful cover letter. Assessment complete. Then an automated rejection lands eleven minutes later, with no human name attached. The rejection is not what stings. The eleven minutes is.
This kind of experience is becoming more common across UK hiring, and for SMEs competing for the same talent as larger, better-resourced employers, it is a quiet reputational risk worth taking seriously.
Scope note: This post offers regulatory context, not legal advice, and is not a compliance checklist. UK data protection law changed materially on 5 February 2026 when section 80 of the Data (Use and Access) Act 2025 replaced UK GDPR Article 22 with new Articles 22A to 22D. Any employer using AI in hiring should also consider lawful basis, transparency, data protection impact assessments, accuracy, bias monitoring, accessibility and candidate rights. A qualified UK data protection or employment law specialist should advise on your specific setup.
Candidate experience is the sum of every interaction a person has with your company from the moment they see your job advert to the moment they either join or walk away. In 2020 that meant a job ad, an application form, a phone screen, an interview, and a decision. In 2026, more of those touchpoints are AI-mediated. The job ad may be drafted with a large language model. The application form may be parsed by a screening tool. The first interview is sometimes a one-way video assessment scored partly by an algorithm. The rejection, if it comes, may arrive automatically.
None of this is inherently bad. Automation done well can be faster and more consistent, though it is not automatically fairer, that depends on the data, the design, the validation and the human governance around it. Automation done badly can quietly repel the exact candidates you want to attract. Most UK SMEs have not yet stress-tested which side of that line they sit on.
Here is the paradox. Candidates want faster processes. They also want to feel seen. AI can help deliver the first. It often struggles with the second.
A 2025 Zinc survey of 1,000 UK HR and talent professionals reported that 73% of respondents used AI somewhere in recruitment, and 71% believed automation removes personalisation from the hiring process. Greenhouse’s 2025 AI in Hiring Report, which surveyed 4,136 job seekers, recruiters and hiring managers across the US, UK, Ireland and Germany, found that 70% of hiring managers trust AI to make faster and better decisions, while only 8% of candidates call the process fair.
Meanwhile, the CIPD Resourcing and Talent Planning 2024 report, based on 1,016 UK people professionals surveyed by YouGov, found 27% said new starters “always, mostly or sometimes” fail to turn up on day one, and 41% said new hires resign within twelve weeks on the same qualified basis. The figures point to a real onboarding and retention challenge; they do not prove that AI or poor candidate experience caused those outcomes on their own. Poor communication and unclear expectations often contribute. This is the same territory I explored in why new hires quit within 90 days, and the AI layer may now be accelerating the problem.
Used well, AI removes some of the parts of the process candidates already hated. Sifting hundreds of CVs by hand meant many strong applications never received a proper read. AI screening, if it is audited for bias, layered with human review, and measured for false negatives, can widen who gets seriously considered rather than narrow it.
Scheduling is a quieter win. Anyone who has played diary tennis for a week to book a thirty-minute interview knows the frustration. Automated scheduling tools cut that to a single click.
Adverts drafted with AI support can save real time and help flag jargon or exclusionary phrasing. Every advert still needs a human review, though: one 2024 study of 1,439 adverts by Develop Diverse (a vendor of inclusive-writing software) found GPT-4-generated adverts scored worse on inclusivity than human-written ones on that sample. Support the drafter; do not hand them the pen. I explore the wider attraction question in AI-powered attraction and employer branding.
Rejection is one of the moments most likely to shape how a candidate talks about your company afterwards. An automated no, sent within minutes with no named human attached, sends a signal that no one really looked. A short, personalised note from a recruiter is not a hard fix, and it protects the reputation of a company that keeps hiring from the same community.
Interviews are the second moment. AI-scored one-way video interviews are efficient, but they strip candidates of the ability to ask questions, read the room, or decide whether they actually want the job. If you use them, use them early in the process, not as the only interview. I go deeper into this in AI interviews and assessments for UK SMEs.
Offer stage is the third. A live human phone call at offer stage tends to convert better than an emailed offer letter alone. Small effort, meaningful lift.
The rule I give SME leaders is straightforward. Automate the volume. Humanise the moments that matter.
Volume tasks are advert drafting, CV sifting, initial scheduling, status updates and structured assessment scoring. These benefit from AI when the outputs are audited. Moments that matter are first contact, interview conversation, offer and rejection. These benefit from a named, accountable human.
If a candidate reaches the end of your process without recognising any human as accountable for the decision, treat that as a signal to review the process. Then measure it: time to first response, time to final decision, candidate withdrawal rate at each stage, offer acceptance rate, completion rate, and a periodic feedback score from a mixed sample of successful and rejected candidates.
It depends on the decision, the data used, and the safeguards in place. On 5 February 2026, section 80 of the Data (Use and Access) Act 2025 replaced UK GDPR Article 22 with new Articles 22A to 22D. A solely automated significant decision, one made with no meaningful human involvement that has a legal or similarly significant effect, is now generally permitted where the Article 22C safeguards apply: transparent information about the processing, an opportunity to make representations, human intervention, and a right to contest the decision. Where special-category data is involved, the more restrictive Article 22B applies. A nominal human “rubber stamp” is not meaningful human involvement; the reviewer needs the authority and competence to actually change the outcome. This is general context, not legal advice.
Monitor a handful of measurable metrics rather than relying on gut feel: time to first response, time to final decision, candidate withdrawal rate at each stage, offer acceptance rate, and structured feedback scores from a varied sample of successful and rejected candidates. Look for patterns over time, not a single figure in isolation.
Personalised, named rejections at final-round stage. The cost is low, and it protects your reputation in the specific talent market you keep hiring from.
About the author Sabiha is a Talent Acquisition Director, speaker and author with over 16 years of international hiring experience across the UK, Dubai, South Africa and Malaysia. She has advised 300+ businesses on hiring, retention and AI-enabled workforce strategy, and was shortlisted for Best Career Coach UK by the Career Development Institute. Her forthcoming book, How to Use AI to Win Talent and Retain People (Trotman, CIPD-aligned, Autumn 2026), sets out a practical framework for UK employers navigating AI in recruitment and retention.
Global Talent. Ethical AI. Strategic Hiring. Sustainable Retention.
