As employers increasingly rely on pre-hire assessments to vet frontline talent, a new study from HiringBranch cautions against a common practice: reporting soft skills like empathy, active listening, and reassurance as isolated scores. The research, discussed in the latest episode of the podcast "You Should Know," suggests that this approach produces only moderate correlation with human judgment, while a combined proprietary model yields much stronger predictions of on-the-job performance.
The episode, titled "Assessing Skills One at a Time Is Costing You Better Hires," was published on August 26, 2026, and hosted by William Tincup. It features Assaf Bar-Moshe, Chief of Research and Development Officer at HiringBranch, a Montreal-based assessment company. The conversation arrives as businesses rethink how they measure customer service representatives, sales agents, and retail associates against the realities of live customer interactions.
Bar-Moshe explained that HiringBranch's assessment design uses open-ended voice and writing prompts to evaluate candidates. The company measures four pillars of customer service: acknowledgment, reassurance through positive language, empathy, and active listening. These are assessed through conversation flows and scenario-based tasks that are calibrated per client, region, and role.
The research found that when these skills are scored separately, the correlation with human annotators is only moderate. However, when combined into a single model, the correlation strengthens significantly. Bar-Moshe illustrated the point with a retail example: "If a candidate can express empathy, but is unable to solve the issue correctly or to comprehend the issue correctly or to reassure the customer, then this empathy is nice, but it's actually useless."
This linguistic approach, rather than personality-based testing, is central to HiringBranch's methodology. Bar-Moshe, a trained linguist, described the method as a "sociopragmatic analysis of the words that the candidate is actually saying." The company's team of IO psychologists and linguists uses years of textual data to build machine learning models that predict empathy, acknowledgment, and related skills. These predictions are then validated against on-the-job performance months after hire.
Calibration is key, as scoring weights differ by client, industry, and even geography. For example, regional variations across markets like Vancouver, Toronto, and Montreal can shape scoring for the same role. Bar-Moshe also previewed a self-serve capability in development that would allow hiring managers to build assessments from a library of conversation flows and skills, reducing reliance on weak or generic job descriptions.
The full study will be available under the AI research tab on the HiringBranch website. The implications for employers are significant: by moving away from isolated soft skills scoring, companies may improve their ability to select candidates who can effectively handle customer interactions, ultimately reducing turnover and enhancing service quality. This research challenges long-held assumptions in pre-hire evaluation and could reshape how frontline hiring is conducted across industries.
For more insights, the episode is available on the You Should Know Podcast page.
