Estimating causal effects of customer satisfaction on downstream metrics in a multi-queue contact center

TL;DR

This study uses an instrumental-variable approach to accurately estimate how customer satisfaction in airline contact centers causally affects downstream business metrics, addressing biases from observational data.

Contribution

It introduces a novel instrumental-variable methodology tailored for multi-queue contact centers, enabling causal inference from observational data.

Findings

Instrumental-variable approach reveals significant causal effects.

Spurious correlations can bias traditional estimates.

Methodology aids operational decision-making in contact centers.

Abstract

Contact centers are crucial in shaping customer experience, especially in industries like airlines where they significantly influence brand perception and satisfaction. Despite their importance, the effect of contact center improvements on business metrics remains uncertain, complicating investment decisions and often leading to insufficient resource allocation. This paper employs an instrumental-variable approach to estimate the causal effect of customer service interactions at the contact center of LATAM airlines on downstream metrics. Leveraging observational data and the examiner design, we identify causal effects through the quasi-random assignment of agents to calls, accounting for the multi-queue structure and agent certification heterogeneity. Our empirical results highlight the necessity of an instrumental variable approach to accurately estimate causal effects in contact centers, revealing substantial biases from spurious correlations. This methodology provides managers with tools to estimate the impact of call satisfaction on key business metrics, offering valuable insights to solve operational trade-offs of call centers.