An Empirically Developed Method to Aid Decisions on Architectural Technical Debt Refactoring: AnaConDebt
Paper i proceeding, 2016

Architectural Technical Debt is regarded as sub-optimal architectural solutions that need to be refactored in order to avoid the payment of a costly interest in the future. However, decisions on if and when to refactor architecture are extremely important and difficult to take, since changing software at the architectural level is quite expensive. Therefore it is important, for software organizations, to have methods and tools that aid architects and managers to understand if Architecture Technical Debt will generate a costly and growing interest to be paid or not. Current knowledge, especially empirically developed and evaluated, is quite scarce. In this paper we developed and evaluated a method, AnaConDebt, by analyzing, together with several practitioners, 12 existing cases of Architecture Debt in 6 companies. The method has been refined several times in order to be useful and effective in practice. We also report the evaluation of the method with a final case, for which we present anonymized results and subsequent refactoring decisions. The method consists of several components that need to be analyzed, combining the theoretical Technical Debt framework and the practical experience of the practitioners, in order to identify the key factors involved in the growth of interest. The output of the method shows summarized indicators that visualizes the factors in a useful way for the stakeholders. This analysis aids the practitioners in deciding on if and when to refactor Architectural Technical Debt items. The method has been evaluated and has been proven useful to support the architects into systematically analyze and decide upon a case.

decision making



design research


Architectural Technical Debt


Antonio Martini

Chalmers, Data- och informationsteknik, Software Engineering

Jan Bosch

Chalmers, Data- och informationsteknik, Software Engineering

Proceedings - International Conference on Software Engineering

02705257 (ISSN)

978-1-4503-4205-6 (ISBN)







Mer information

Senast uppdaterat