Effort Estimation in Agile and DevOps: A Systematic Literature Review of Stakeholder Dissatisfaction

Effort estimation is a critical aspect of software project management, directly influencing the budget, schedule, and resource allocation throughout the software development life cycle. Accurate estimation plays a central role in aligning team capabilities with stakeholder expectations, ensuring timely delivery and efficient use of resources. Despite the adoption of Agile and DevOps methodologies, which emphasize flexibility, rapid iterations and continuous integration, estimation inaccuracies remain a problematic area. These inaccuracies often result in project delays, cost overruns, and scope creep, ultimately leading to stakeholder dissatisfaction and a breakdown of trust in the development process. To investigate these challenges, a Systematic Literature Review (SLR) was conducted, surrounding 1,088 publications retrieved from six major digital libraries: Scopus (201), ScienceDirect (155), Springer (212), Google Scholar (145), IEEE (185), and ACM (190). This research examines the main drivers of estimation challenges in Agile and DevOps environments. Key findings indicate that human-centric issues (cognitive biases, lack of experience), technical issues (fluctuating requirements, lack of adequate history), and process inefficiencies (communication breakdown, poor planning) exert considerable influences on estimation accuracy. To address these challenges, the study categorizes and analyzes a range of estimation techniques, including machine learning models, expert-calibrated hybrid approaches, analogical estimation methods, and process calibration practices such as retrospectives. Findings reveal that a combination of data-driven tools and stakeholder-aligned strategies offers the most promise in narrowing the expectation experience gap. This research provides practitioners with practical recommendations and lays the ground for further study in developing estimation techniques to mitigate the expectation-experience gap.