TL;DR
- CVs reveal career history, not future performance. Worse, CV-trained hiring models amplify existing bias, which systematically filters out strong candidates.
- Predictive assessment, on the other hand, measures the behavioural competencies and cognitive ability that drive job performance in the real world.
- Sapia.ai’s conversation-based approach goes further than traditional predictive index (PI) behavioural assessment tools by scoring what candidates say and how they say it.
- Over 80% of candidates hired by Sapia.ai customers come from the “Yes” or “Maybe” shortlist that’s generated by our predictive index (PI) assessment.
Does your hiring team still use CVs as its first filter in the hiring process? On the surface, that makes sense. CVs are structured, easy to scan, and universally understood. But…
CVs measure the wrong things, which limits your hiring team’s decision-making abilities.
Think about it: A CV tells you where someone has been, not where they’re going. A predictive index assessment tells you the opposite. Once you add them to your interview process, you’ll acquire the data points you need to assess personality traits, evaluate skill sets, see pattern insights, and predict job performance within your work environment with greater accuracy.
Why CVs are a poor predictor of performance
We understand why you want to screen CVs. Application volumes are high, you’re low on time, and CVs give you something to anchor decisions to. But that anchor is mostly noise.
At Sapia.ai, we analysed approximately 13,000 CVs for a large sales organisation over a five-year period. There was a negligible correlation between a candidate’s CV and their on-the-job performance. Put simply, the CV predicted who got hired, not who succeeded.
This is the core problem. When you screen CVs, you replicate hiring decisions of the past rather than build a model of future performance. If those past decisions carried bias? You carry it forward, which not only leads to poor-fit hires and high turnover, but also compliance issues.
Amazon learned this the hard way. The company built an AI-based CV screening tool that analysed the CVs of previous hires, which skewed male. As such, the system penalised female candidates because they didn’t match the “profile.” In other words, the AI worked perfectly, but CVs were a flaw in the training data, so the AI optimised for the wrong target.
CVs also encode signals that have nothing to do with job performance. Examples include the prestige of a university, gaps in employment, and the visual formatting of documents.
These triggers activate unconscious bias before a recruiter reads a single line about a person. Because of this, the system filters out strong candidates as their backgrounds don’t match those of previous employees, while weaker candidates advance because their CVs look familiar.
What predictive assessment actually measures
Where CVs look backwards, a well-designed predictive assessment looks forward.
The predictive index measures behavioural competencies, cognitive traits, and communication skills through structured, job-relevant questions that are scored against validated rubrics.
Ultimately, the questions reveal how assessment takers think, handle pressure, collaborate, and approach problems. The assessment results have nothing to do with the institution the job seeker attended or the previous positions they held.
This matters because the qualities that drive success, like how someone handles a difficult customer, approaches a new problem, or performs under pressure, don’t show up on a CV. They show up when you ask the right questions and measure the answers properly.
It’s worth distinguishing Sapia.ai’s approach from both generic predictive index behavioural assessments and predictive index cognitive assessments. These tools rely on self-report questionnaires that ask candidates to rate themselves. While said tools have a place, they rely on each candidate’s self-concept, not on demonstrated behaviour. A candidate who scores high on a self-rated “resilience” scale may or may not behave resiliently under pressure.
Sapia.ai’s predictive assessment takes a different approach. Candidates respond to structured interview questions in a conversational chat format. Then, NLP models score the content against competency rubrics. Basically, we judge candidates on what they say and how they say it. This gives hiring teams a more accurate and defensible basis for shortlist decisions.
How predictive hiring assessments reduce bias
Predictive assessments help hiring teams build stronger, fairer shortlists. (That way, your company doesn’t create an “Amazon” situation for itself, and engages the best talent.)
Blind first-pass scoring removes name, photo, and CV details from the initial review. This allows recruiters to assess evidence before they know about a candidate’s background.
Additionally, every candidate answers the same questions, which are then scored against the same criteria to remove the interviewer variability that makes unstructured screening unreliable.
Tools like Sapia.ai also monitor for adverse impact. We test our models across gender and racial groups and publish the results. This is one reason why our platform scores 9.2 out of 10 for candidate satisfaction across all demographics.
Another reason why our customer satisfaction scores are so high is that we follow our own FAIR framework, which ensures AI hiring tools have four primary traits: they’re unbiased, valid, explainable, and inclusive. You can trust Sapia.ai’s PI results are fair and accurate.
What this looks like in practice: Woolworths
Woolworths, Australia’s largest private employer, faced a difficult challenge. During COVID, the supermarket chain needed to hire 27,000 team members in less than ten weeks.
Their existing process, which relied on manual CV screening followed by video interviews, couldn’t scale to the necessary volume. So, Woolworths partnered with Sapia.ai to build a fully automated, chat-based smart interview that every retail candidate could complete on their own.
Shortlisted candidates were then automatically progressed to a self-serve video interview. This completely removed the scheduling burden from hiring managers.
As a result, Woolworths filled a significant portion of their open roles within 24 hours of them going live. But the thing is, this isn’t an isolated outcome. At Sapia.ai, we see it all the time.
Validity studies across Sapia.ai’s customer base show strong correlations between assessment scores and both retention and performance outcomes across retail, contact centres, graduate hiring, and healthcare. Over 80% of candidates hired by Sapia.ai customers consistently come from the “Yes” or “Maybe” recommendations generated by our predictive assessments.
The real cost of getting this wrong
The real cost of CV-based screening is the talent you never see.
The candidate who spent five years raising a family and then re-entered the workforce. The career changer whose skills translate perfectly but whose job titles don’t match. The high performer who never attended a prestigious university because they couldn’t afford it.
CV screening filters out these people before they can demonstrate their skills. Predictive assessment allows them to demonstrate that their abilities match your job description—without forcing them to give detailed explanations. It also gives your organisation access to a wider talent pool.
Hiring teams that make the shift often find the candidates who looked unremarkable on paper are the strongest performers. The data just wasn’t there to surface them before.
If you want to find top candidates with the right personality, behavioural drive, and skill sets, book a demo of Sapia.ai to see how predictive assessment works in a real hiring workflow.
FAQs about predictive assessment in the hiring process
A predictive assessment evaluates behavioural competencies and cognitive ability through structured responses that are then scored based on predefined criteria. A CV screen filters on career history. Put simply, one predicts future performance while the other reproduces past hiring patterns.
By removing CV details from the first-pass review, standardising questions, and scoring every candidate against the same validated criteria, you can base hiring decisions on evidence instead of background signals. This naturally removes unconscious bias.
Unlike a predictive index behavioural assessment or predictive index cognitive assessment, which use self-report scales, Sapia.ai’s assessment evaluates how candidates respond to structured interview questions. These answers are then scored by NLP models against competency rubrics.
Yes. Platforms like Sapia.ai are purpose-built for volume hiring. In fact, our solution makes it easier to hire talent at scale because we remove much of the admin work while maintaining quality.
Users typically see reduced time to hire, stronger quality of hire, and more diverse shortlists in the first hiring cycle. For example, Sapia.ai customers report up to 50% reduction in time to hire and an 89% reduction in employee turnover. Book a demo to see our platform for yourself.
About Author
