How a $1,800 Repair Became a $3,000 Problem
When services don’t adapt, costs hit customers first and come back to the business
Over the December holidays, my car was broken into in New York City.
The rear glass was shattered. The car was exposed.
That context matters.
In NYC, a damaged car cannot just sit:
Garages will not take it
Street parking exposes it to the weather and theft
Winter turns a contained issue into something bigger, fast
Speed was not a convenience. It was damage containment, and delay was the cost.
This pattern shows up in services where conditions change faster than systems can adapt. When that happens, outcomes start to break.¹ ⁵
What Was Actually Happening
From the moment the claim was filed, the conditions looked like this:
No safe place to store the car
Active exposure increased damage risk
Holiday timing limited repair availability
At the same time, the system operated as designed—optimized for standard flow, not real-world variability:
A mobile-first flow built for delay, not urgency
Approved vendors with no near-term availability
A nearby dealer that could fix it immediately—but was blocked by policy
No case ownership, so every interaction resets the situation
The issue wasn’t the claim.
It was a system operating outside the conditions it was built for.
When real-world conditions fall outside a system’s design envelope, performance degrades predictably.¹ ⁵
Why It Escalated
Emotion isn’t noise.
It’s a real-time signal that the situation is no longer manageable.
People continuously appraise risk, control, and their ability to cope.² ³
As the car sat exposed, urgency increased, and confidence dropped.
There was no clear path forward.
No control over what would happen next.
The situation was getting worse over time.
This wasn’t just uncertainty. It was a loss of coping ability.
When people can’t see a credible path to resolution, they act to reduce risk.⁴
What Happened Next
The behavior was predictable. And measurable.
Customer behavior:
Repeated calls to establish urgency
Attempts to find faster solutions
Escalation as risk became visible
System behavior:
Adherence to vendor and policy constraints
Fragmented handling across agents
Delayed approvals while exposure continued
Each interaction restarted the case.
Context had to be rebuilt and reinterpreted every time.
This wasn’t resistance.
It was compensation for a system that couldn’t hold together.
Behavior shifts when capability, opportunity, or motivation are constrained.⁶
What the Delay Actually Cost
Original estimate:
~$1,805
Cost of delay:
~$2,500 to $3,400
More than the repair itself.
Where it showed up:
Customer time and lost wages
Rental extension
Duplicate inspections
Vendor churn and towing
Repeated handling and coordination
System overhead
This excludes additional damage from extended exposure.
None of these costs was part of the original problem. They were introduced by the system’s inability to adapt.
This is the cost of running a context-dependent need through a system designed for standard conditions.
The system protected the process.
The cost showed up everywhere else: operationally, financially, and experientially.
The Miss
Most teams diagnose this as a behavior problem:
The customer escalated too quickly
The process wasn’t followed
That misses the mechanism.
Escalation is not the problem. It’s the signal.
When systems cannot adapt to changing conditions:
Uncertainty rises
Control collapses
People compensate through workarounds and escalation.
What looks like friction is the system revealing its limits.
Where AI Actually Helps
This is not primarily a data problem. It’s a signal detection problem.
The data is already there:
Repeat contact
Rising urgency
Vendor mismatch
Incomplete resolution steps
The system isn’t reading the signals early enough to act.
Escalation is not an exception; it is a detectable pattern.⁷
AI’s role is to:
Detect when cases exit expected conditions
Identify loss of control and rising uncertainty
Predict escalation before breakdown⁸
Trigger earlier intervention and ownership
This shifts optimization from managing workflows to managing system stability.
What You Can Do Differently
These changes align systems with how people actually behave under uncertainty, drawing on research in service design, behavioral science, and system performance. (see Additional Reading)
Detect non-routine conditions early
(exposure risk, time sensitivity, vendor mismatch → measurable signals)
→ e.g., flag any case with 2+ contacts and no scheduled resolution within 48 hoursStabilize context through ownership
(reduce rework, improve coordination, increase accountability)
→ e.g., assign one case owner once escalation signals appearEnable controlled exceptions
(optimize total cost, not local compliance)
→ e.g., allow expanded vendor use when approved options exceed wait thresholdsReduce uncertainty immediately
→(clear next steps, timing, fallback)
e.g., provide a same-day plan with timeline and fallbackMeasure total system outcomes
(not just process adherence, but downstream cost and experience)
→ e.g., track cost of delay alongside claim cost
The Real Opportunity
This isn’t specific to insurance.
It shows up anywhere a system encounters real-world variability:
Healthcare
Financial services
Customer support
When conditions change, and a service doesn’t adapt:
emotion escalates
behavior shifts
outcomes degrade
These outcomes are not driven by people. They are produced by the system.
This is a system design problem.
The Point
We don’t fix behavior. We fix the conditions that drive it.
hello@stickwithglue.com
Methods note. This article uses established frameworks from health systems, emotion theory, uncertainty, and behavior change to explain how service conditions drive outcomes.¹–⁶ These references inform the framing and mechanisms described here; the cost figures and operational details are drawn from a single, real-world claim experience rather than a formal empirical study.¹–⁶
References
Donabedian, A. (1966). Evaluating the quality of medical care. Milbank Memorial Fund Quarterly, 44(3), 166–203.
Lazarus, R. S. (1991). Emotion and adaptation. Oxford University Press.
Scherer, K. R. (2005). What are emotions? And how can they be measured? Social Science Information, 44(4), 695–729.
Carleton, R. N. (2016). Fear of the unknown: One fear to rule them all? Journal of Anxiety Disorders, 41, 5–21.
World Health Organization. (2021). Health system performance assessment: A framework for policy analysis. World Health Organization.
Michie, S., van Stralen, M. M., & West, R. (2011). The behaviour change wheel: A new method for characterising and designing behaviour change interventions. Implementation Science, 6, 42.
Anirudh K., et al. (2020). Customer Support Ticket Escalation Prediction using Feature Engineering. arXiv:2010.06145.
Zhao Y., et al. (2023). Developing an Artificial Intelligence-Guided Signal Detection System. Frontiers in Pharmacology.
Additional Reading
The works below are relevant to the broader themes (claims friction, self‑service failure, escalation, and system design) but are not directly cited in the main text above.
Not directly cited in the text, but relevant to the themes and empirical backdrop:
Dahle, L. H. (2016). Designing for people in crisis: Service design for an emergency room. Department of Product Design, Norwegian University of Science and Technology (NTNU).
Manderson, K., Taylor, N. F., Lewis, A., & Harding, K. E. (2025). Service-level interventions to reduce waiting time in outpatient and community health settings may be sustained: A systematic review. BMJ Open Quality, 14(1).
McKinsey & Company. (2016). The growth engine: Superior customer experience in insurance.
In2. (2024). Inefficiencies in insurance claim management and the legacy systems dilemma.
Insurance Information Institute. (2024). Auto insurance claims and loss severity.
Buell, R. W., Campbell, D., & Frei, F. X. (2025). Are self-service customers satisfied or stuck? Production and Operations Management.
Moon, Y., & Frei, F. X. (2000). Exploding the self-service myth. Harvard Business Review, 78(3), 26–27.
Gallup. (2014). Why great managers are so rare. Gallup Workplace.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Tversky, A., & Kahneman, D. (1991). Loss aversion in riskless choice: A reference-dependent model. Quarterly Journal of Economics, 106(4), 1039–1061.
Weick, K. E. (1995). Sensemaking in organizations. Sage Publications.
Batalden, P. B., & Davidoff, F. (2007). What is “quality improvement” and how can it transform healthcare? Quality and Safety in Health Care, 16(1), 2–3.
Solar, O., & Irwin, A. (2010). A conceptual framework for action on the social determinants of health. World Health Organization.
Robotham, D., et al. (2016). Appointment reminder systems are effective but not optimal: Results of a systematic review and evidence synthesis. BMC Health Services Research, 16, 394.
Scherer, K. R. (2013). Driving the emotion process: The appraisal component. In M. D. Robinson, E. R. Watkins, & E. Harmon-Jones (Eds.), Handbook of cognition and emotion (Chapter 12). Oxford University Press.
West, R. (2020). A brief introduction to the COM‑B model of behaviour and the PRIME theory of motivation. Prevention Collaborative / University College London.