What is the Long-Term Value of Reliability?

2026-06-09

AI Summary

This study investigates the long-term impact of service reliability issues—such as latency—on key business metrics. To capture the dynamic interactions between customers and the system, the authors formulate a Markov decision process model, defining the target quantity as the marginal policy effect under varying average delay rates. Under a sequential unconfoundedness assumption grounded in observed order-level features, they establish conditions for identifying and estimating long-term causal effects. The work innovatively adapts the marginal policy effect framework to service reliability evaluation, introducing a novel sequential unconfoundedness criterion based on observable order characteristics. By integrating covariate balancing algorithms, the authors develop the first identifiable and estimable quantitative approach for assessing the long-term business consequences of service deficiencies, thereby offering precise decision support for reliability investments.

Abstract

We describe Chronos LTV, a system to measure the long-term impact of delays and other service defects on key business metrics. We use Markov decision processes to model customer interactions over time, and formalize our target estimand as the marginal policy effect with respect to moving the average delay rate. Given this setup, we show that we can identify long-term effects under a sequential unconfoundedness assumption where delays are as good as random given observed order characteristics; and can estimate these effects using a simple covariate-balancing algorithm.

Problem

• long-term value
• reliability
• service delays
• business metrics
• customer interactions

Innovation

• Markov decision processes
• marginal policy effect
• sequential unconfoundedness
• covariate balancing
• long-term value