Peyton Fry: Fixing the Front Door of Healthcare

Peyton Fry is the founder of Glass Raven, a consulting firm that helps healthcare systems improve patient access and call center operations. He started answering phones at Best Buy at sixteen, triaged telehealth calls while completing his graduate degree at Texas A&M, and later moved into patient access at Houston Methodist. The path gave him direct experience of the pressures faced by agents, supervisors, analysts, and executives within a call center. It also shaped his view that performance improves when teams track fewer measures and connect each one to a specific decision.

Many call centers operate with dashboards that stretch from columns A through Z. Managers review them daily, yet few of the numbers change what anyone does. Fry strips the dashboard back to zero and tests every metric against one question: which decision depends on this number? Any measure without a clear answer comes out.

The exercise usually leaves the service level at the center of the operating model. Service level measures the share of calls answered within a defined time and provides leaders with a basis for staffing decisions. Staffing accounts for most of the cost, so the metric directly links performance to budget. Measures such as abandonment rate generally improve once the team has enough people available at the right times. A single target can therefore explain several outcomes that were previously scattered across a large dashboard.

Accurate staffing requires a deeper view of demand than standard online calculators provide. Call volumes shift across the year, between days of the week, and throughout each day. Fry models all three layers because small forecasting errors can lead an organization to hire ten people when four would have been enough. He describes the gap as a $250,000 decision.

Weekly demand patterns create another common mistake. Monday is frequently the busiest day, and staffing the entire operation around its peak can leave 15 people underused for much of the rest of the week. Fry plans capacity across the full week. The team may miss its target during part of Monday and still deliver the required service level across the broader period. This gives leaders a more efficient balance between customer access and labor cost.

Average handle time feeds into the same calculation. Cutting 45 seconds from each call can remove the need for 20 additional hires. Leaders can compare the cost of training, better workflows, or improved systems with the cost of adding headcount.

Fry also assigns each measure to the person who can influence it. Calls answered per day tell managers very little about agent performance because incoming demand determines the volume. Handle time and schedule adherence reflect behaviors that agents can control. At one client, the team narrowed agent accountability to a small group of controllable measures, raised its service-level target from 70/90 to 80/30, and exceeded the new standard. Some employees left under the tighter expectations, several positions remained unfilled, and the organization improved performance while reducing staffing costs without increasing its budget.

The same principle shapes Fry’s work with clinicians and decision teams. Doctors usually assess one patient at a time, while analysts examine patterns across a larger population. Collaboration improves when the analyst understands the clinician’s constraints, reduces the administrative burden, and takes responsibility for the supporting work.

A small set of metrics tied to staffing, training, and workflow decisions gives a team an operating system. A dashboard filled with disconnected figures gives it more information but less direction. Clarity wins over coverage.

What is preventing your AI investments from delivering stronger business results?

The Operational Research Society's flagship annual conference brings together decision scientists, operational researchers, analytics leaders, and industry practitioners to explore how analytical methods are improving decisions across business, government, and public services. The 2026 theme, "From Data to Decisions," focuses on translating data into operational action through optimization, simulation, analytics, and decision science. With technical sessions, applied case studies, and practitioner discussions spanning multiple industries, OR68 offers a valuable perspective for organizations looking to strengthen the science behind enterprise decision-making. Details →

Hosted by the Italian Association of Operations Research, ODS2026 brings together researchers, optimization specialists, and decision science practitioners to discuss advances in mathematical optimization, prescriptive analytics, decision support, and operational research. The program spans optimization algorithms, uncertainty modeling, simulation, and real-world applications across logistics, healthcare, manufacturing, and energy. Details →

This paper argues that AI governance should be treated as a dynamic decision problem and introduces a Bayesian framework that continuously adjusts delegated decision authority based on evidence quality, uncertainty, and organizational risk, replacing fixed confidence thresholds with adaptive governance policies. Researchers also benchmark the approach against five common governance strategies, showing that adaptive delegation performs more consistently as AI capabilities evolve. Read →

Carl Spetzler, Fellow of the Society of Decision Professionals and Founding Member of the DecideWise community, argues that Decision Intelligence needs to be understood as a continuum. On one end sit high-volume, automatable decisions, pricing, fraud detection, and routing, where data scientists and operations researchers already excel at scale and speed. On the other side, rare, high-stakes strategic calls where Decision Quality practitioners bring structured framing, explicit values, and reasoning under genuine uncertainty. In between lies what Spetzler calls the "muddled middle", the vast volume of daily decisions made by managers and project leads, neither routine enough to automate nor consequential enough to warrant formal decision analysis. His argument is that AI is compressing this middle from both directions, automating what used to require judgment, while making structured decision frameworks accessible to people who'd never have had access to a facilitated workshop. The risk is that the two communities at the two ends of the continuum, data science and decision analysis, remain largely unaware of each other's traditions. Read More →

Warren Powell, Co-Founder of Optimal Dynamics, joins Adam DeJans Jr. of Gurobi Optimization, along with hosts Vijay Mehrotra and Michael Watson, for a conversation about the mechanics of deciding well under uncertainty. Powell's Sequential Decision Analytics framework has shaped how supply chain and energy systems handle sequences of decisions rather than one-off choices. He and DeJans discuss why so much of decision-making under uncertainty gets treated as a data problem when it's really a modeling-and-state problem: understanding what state you're actually in, what decision you're actually making, and how uncertainty evolves after you make it. Both guests argue that optimization has remained too long in the hands of specialists, and the real opportunity lies in giving non-experts a way to engage with sequential decisions without needing a PhD to frame the problem correctly.