Your Gut Feeling Isn’t a Hiring Strategy

Three ways evidence-based recruitment and domain-specific AI can help recruiters make stronger, more consistent decisions.

The interview went brilliantly. The candidate was persuasive, personable and apparently perfect. Three months later, the performance tells a different story.

Recruiters know the gap between making a good impression and doing a good job. Evidence-based recruitment tackles that gap by asking a more useful question: what information actually helps us predict success in this role?

Domain-specific AI can support that approach. Here, that means AI developed or configured for a defined recruitment task and evaluated in the relevant job context. The label alone proves nothing: its value depends on the quality of the evidence, the assessment design and the checks around its use.

For recruiters, three benefits are worth pursuing.

1. A stronger case for who makes the shortlist

A polished resume tells you what someone claims to have done. A well-designed assessment helps you investigate what they can do in the role you need to fill.

The research supports taking structure seriously. A major re-analysis by Sackett and colleagues, published in 2022, found that structured interviews ranked highest among the selection methods examined for predicting job performance. It also revised many earlier validity estimates downwards: useful prediction is possible, but certainty is not.[1]

The practical lesson is to define success before assessing candidates. For a team leader, that might include allocating work, handling difficult conversations and maintaining service quality. Interview questions and scoring criteria should then examine those requirements.

AI may help organize relevant evidence or identify patterns for recruiters to investigate. Any claim that its scores predict performance needs validation for the intended use, rather than an assumption that a specialist model must be accurate.

The recruiter’s gain: a shortlist supported by reasons you can examine, discuss, and test against subsequent performance.

2. More consistent comparisons—and clearer explanations

When every interviewer follows a different conversation, candidates can end up facing different tests. One is assessed on technical judgement; another on confidence, or conversational chemistry.

The US Office of Personnel Management describes structured interviews as using predetermined questions and common rating standards. Its guidance explains that greater structure improves agreement between assessors and that questions should reflect competencies identified through job analysis.[2]

That gives recruiters a practical foundation: agree the criteria, collect comparable evidence, and record how each decision was reached.

An AI tool can support the workflow—for example, by arranging evidence against those criteria but its recommendations still need scrutiny. Consistency alone does not establish fairness. A system can apply an unsuitable rule consistently, and historical performance ratings can contain human bias.

NIST’s AI Risk Management Framework calls for contextual evaluation, documented validity and reliability, assessment of harmful bias, and defined human oversight.[3]

The recruiter’s gain: a more transparent basis for comparing applicants and explaining recommendations to hiring managers, with a record that can be reviewed when something looks wrong.

3. A hiring process that can learn after the offer

Filling a vacancy is a milestone. Finding out whether the assessment identified someone who performs well is where the learning begins.

This is particularly relevant to The Cambridge Don’s approach. Its services page describes Talent Chaser as linking a Performance Profile Questionnaire with performance appraisal ratings for specific job positions, using that information to update recruitment screens.[4]

That connection creates an opportunity to examine whether pre-hire assessments relate to later performance. It also prompts better conversations between recruiters and managers: Did the role’s requirements change? Were expectations clear? Did the new employee receive appropriate support?

Those questions matter because performance reflects more than the person recruited. Management, training, and working conditions also affect results. A useful feedback process examines those influences rather than treating every disappointing outcome as a screening failure.

Updating a model is not proof of improvement, either. Revised screens should be tested on data not used to build them, with ongoing checks for performance and harmful bias as roles and applicant populations change.[3]

The recruiter’s gain: an opportunity to improve the hiring process using evidence from what happens on the job, rather than repeating assumptions from the last vacancy.

Ask better questions of your hiring technology

Before adopting a tool, ask: What does it measure? What evidence links that measurement to this role? How is it checked for errors and unfair outcomes? Who reviews its recommendations?

Evidence-based recruitment makes those questions central. Domain-specific AI earns its place when it helps answer them convincingly.

Want to examine the evidence behind your hiring process? [Book a free audit with The Cambridge Don](https://thecambridgedon.freakystuff.site/contact).

Sources

1. Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). [Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range](https://pubmed.ncbi.nlm.nih.gov/34968080/). Journal of Applied Psychology, 107(11), 2040–2068. doi:10.1037/apl0000994. Research on selection methods; not a validation of Talent Chaser or AI recruitment products.

2. US Office of Personnel Management. [Structured Interviews](https://www.opm.gov/policy-data-oversight/assessment-and-selection/other-assessment-methods/structured-interviews/). Practical guidance on job-related questions, scoring, and assessor agreement.

3. National Institute of Standards and Technology (2023). [AI Risk Management Framework 1.0: Core](https://airc.nist.gov/airmf-resources/airmf/5-sec-core/). See GOVERN 3.2 and MEASURE 2.5, 2.9, 2.11 and 3.1. Voluntary risk-management guidance; not evidence of a particular product’s effectiveness.

4. The Cambridge Don. [Services: Evidence-Based Recruitment and Performance Appraisal](https://thecambridgedon.freakystuff.site/services). Source for the description of Talent Chaser’s workflow.

Editorial draft. Sources checked 24 October 2026.