Why the Conversations We Don’t Measure Become the Outcomes
We DoHealthcare organizations measure almost everything. Revenue. Utilization. Turnover. Productivity. Patient experience. Clinical outcomes. Documentation timeliness. Cancellations. Complaints. Compliance. What remains fascinating is that the behavioural interactions that serve as leading indicators of those outcomes often remain largely invisible until something goes wrong. A parent raises a concern about their child’s treatment. The clinician becomes defensive. The parent disengages and cancels sessions. Management gets involved.
A supervisor provides appropriate feedback in a way that shuts down employee disclosure. The employee stops raising concerns. Performance deteriorates. Eventually, the employee leaves. By the time these events appear on an executive dashboard, the organization is measuring the consequence. The failure may have occurred several interactions earlier. Healthcare does not have a “soft skills” problem. It has a relational performance problem and, more specifically, a measurement problem.
We Measure the Outcome, Not the Interaction
Healthcare and many other organizations have become increasingly sophisticated at measuring clinical, financial, and operational performance. If anything matters, we tend to define it, establish standards around it, monitor it, and intervene when performance degrades. We do not operate the same way with human interaction.
Everyone agrees listening, empathy and trust all matter. We invest in leadership development, communication training, patient experience, and employee engagement. However we rarely think of interactions between people as a form of performance that requires the same level of rigor described above. When an important interaction occurs under pressure, we have not defined what effective performance looks like, how it can be observed or how we know that it’s improving.
Instead, we often discover the consequences of relational failure through lagging indicators: a complaint, an escalation, disengagement, turnover, avoidable supervisor time, or a breakdown in adherence. What this creates is measurement of the outcome instead of the behavioral interactions that preceded it. We think of this exposure as relational risk: the possibility that an important conversation, decision, or collaboration breaks down in ways that adversely affect clinical or organizational performance.
It does not mean every cancellation or resignation is caused by poor communication. Healthcare outcomes are multiply determined. It means organizations can become more systematic about a variable they have historically left largely to personality, experience, and good intentions.
We Have Confused Relational Capability With Personality
Part of the problem is how we describe people’s relational capability. We describe people as empathetic, collaborative, difficult, compassionate, strong communicators, or natural leaders. We talk about bedside manner and executive presence. These labels may be convenient, but they can obscure the more important question: What does the person actually do?
What happens when a caregiver becomes angry? When an employee challenges a supervisor? When a clinician must communicate something a patient does not want to hear? When a manager must give corrective feedback to someone who becomes defensive?
Performance under those conditions cannot be inferred reliably from whether someone is generally “nice” or scores well on a self report measure. It shows up in behaviors. Relational capability stops being something a person simply has. It becomes something a person does.
Applied behavior analysis (ABA) has something important to contribute here. ABA is often associated with autism treatment, but behavior analysis brings a science built around turning important human behavior into something observable, teachable, measurable, and improvable. That orientation invites a different way of thinking about relational capability.
The Next Performance Frontier
Healthcare already approaches many important areas of performance this way. We would not expect a clinician to master a complex procedure by attending an annual seminar. We define competent performance, create opportunities to practice, observe performance, provide feedback, and assess whether the skill transfers to real work. Yet that is often exactly how we approach difficult human interactions.
We tell people to communicate better. Lead with empathy. Build trust. Listen actively. Then we send them back into high stakes situations and hope those concepts become behavior when the interaction becomes difficult. The opportunity is to treat relational performance with the same seriousness we apply to other forms of performance. That does not mean reducing human relationships to a simplistic score. Context matters. Culture matters. History matters. Power dynamics matter. It means recognizing that how people interact is itself performance, and that the behaviors involved can be observed, practiced and developed rather than leaving them entirely to personality, experience, and good intentions.
AI Changes the Practice Constraint
Until recently, one practical barrier was scale. The conversations people most need to practice are difficult to reproduce safely and repeatedly. An angry caregiver, a defensive employee, a disagreement between clinicians, or a manager delivering difficult feedback cannot simply be summoned whenever someone needs another practice attempt.
Generative AI changes that constraint. It gives us a way to create opportunities for people to practice difficult human interactions repeatedly and safely. Imagine being able to practice a consequential conversation repeatedly before having it in real life, with variation, increasing difficulty, and feedback without placing a patient, caregiver, employee, or colleague inside the practice environment.
AI does not determine what good human interaction means. Human expertise, evidence, ethics, and context still have to define competent performance. Technology makes it possible to create many more opportunities for deliberate practice. The opportunity is not simply to use AI to replace human capability, but instead to help build and scale it.
What ABA Can Bring to Healthcare
Healthcare does not need another discipline to discover that relationships matter. Medicine, nursing, psychology, and other fields have studied communication, therapeutic relationships, empathy, and trust for decades.
ABA can make a different contribution. Behavior analysis brings a scientific tradition of asking questions that organizations now need to ask about relational performance: What does effective behavior actually look like? Under what conditions does it occur? What happens when conditions become difficult? Can it be taught? Can it be practiced with guidance? Does it generalize? And does improvement change meaningful outcomes? Those questions point toward a larger opportunity for healthcare. For decades, organizations have become increasingly sophisticated about clinical performance, financial performance, operational performance, and quality.
Relational performance may be the next frontier. As technology becomes better at performing portions of healthcare’s technical and administrative work, the interactions that remain may become more, not less, consequential. Those interactions are part of what makes healthcare human, and increasingly, they may help define what effective human performance looks like.
The challenge is no longer convincing healthcare that trust matters. It is understanding the behaviors that build or erode trust before their consequences show on a dashboard. Healthcare has spent decades learning to measure what happens after important conversations. It may be time to get much better at what happens inside them.
Selected research informing this article
- Taylor, B. A., LeBlanc, L. A., & Nosik, M. R. (2018). Compassionate care in behavior analytic treatment: Can outcomes be enhanced by attending to relationships with caregivers? Behavior Analysis in Practice.
- EY (2026), “How to redesign work around human skills in the age of AI.”
- Deloitte (2026), Global Human Capital Trends: “Choosing the human advantage.”