Asymmetric Release Planning-Compromising Satisfaction against Dissatisfaction

Maximizing satisfaction from offering features as part of the upcoming release(s) is different from minimizing dissatisfaction gained from not offering features. This asymmetric behavior has never been utilized for product release planning. We study Asymmetric Release Planning (ARP) by accommodating asymmetric feature evaluation. We formulated and solved ARP as a bi-criteria optimization problem. In its essence, it is the search for optimized trade-offs between maximum stakeholder satisfaction and minimum dissatisfaction. Different techniques including a continuous variant of Kano analysis are available to predict the impact on satisfaction and dissatisfaction with a product release from offering or not offering a feature. As a proof of concept, we validated the proposed solution approach called Satisfaction-Dissatisfaction Optimizer (SDO) via a real-world case study project. From running three replications with varying effort capacities, we demonstrate that SDO generates optimized trade-off solutions being (i) of a different value profile and different structure, (ii) superior to the application of random search and heuristics in terms of quality and completeness, and (iii) superior to the usage of manually generated solutions generated from managers of the case study company. A survey with 20 stakeholders evaluated the applicability and usefulness of the generated results.

💡 Research Summary

The paper introduces a novel perspective on software release planning by explicitly distinguishing between two opposing stakeholder reactions: satisfaction gained when a feature is delivered and dissatisfaction incurred when it is omitted. Traditional release planning approaches treat these reactions as a single utility measure, typically focusing on maximizing overall value or customer satisfaction. The authors argue that this conflation overlooks the asymmetric nature of stakeholder expectations: a feature that is highly desired may generate substantial dissatisfaction if absent, while a less‑desired feature may contribute little additional satisfaction when present. To capture this asymmetry, they formulate the Asymmetric Release Planning (ARP) problem as a bi‑criteria 0‑1 integer linear program. The decision variables indicate whether each candidate feature is selected for the upcoming release. Two objective functions are defined: (1) maximize the total expected satisfaction ( \sum_i S_i x_i ) and (2) minimize the total expected dissatisfaction ( \sum_i D_i (1-x_i) ). Here, (S_i) and (D_i) are feature‑specific parameters representing, respectively, the incremental satisfaction if the feature is delivered and the incremental dissatisfaction if it is not. Constraints enforce the overall effort budget, prerequisite relationships among features, and the binary nature of the selection variables.

A central contribution is the method for estimating (S_i) and (D_i) from stakeholder input. The authors extend the classic Kano model, which classifies features into categories such as “must‑be”, “one‑dimensional”, and “attractive”, by introducing a continuous variant. In a standard Kano questionnaire, respondents answer two Likert‑scale questions per feature: one about their reaction if the feature is present and another about their reaction if it is absent. The continuous Kano approach maps these ordinal responses onto a 0‑1 interval, assigns category‑specific weights for satisfaction and dissatisfaction, and aggregates the weighted responses to produce numeric estimates for (S_i) and (D_i). This yields a richer, quantitative representation of the asymmetric impact of each feature.

To solve the bi‑objective model, the authors develop the Satisfaction‑Dissatisfaction Optimizer (SDO). SDO proceeds in three stages: (1) data collection and preprocessing of the continuous Kano questionnaire, (2) computation of the (S_i) and (D_i) parameters, and (3) generation of a Pareto‑optimal frontier using mixed‑integer programming solvers (CPLEX/Gurobi). The frontier is approximated by a hybrid of ε‑constraint and weighted‑sum techniques, systematically varying the relative importance of satisfaction versus dissatisfaction to produce a diverse set of trade‑off solutions. This approach ensures that decision makers can explore a spectrum of release plans ranging from “satisfaction‑focused” to “dissatisfaction‑avoidance‑focused”.

The methodology is validated through a real‑world case study at a Korean software company developing a mobile application. The project involved 30 candidate features and an estimated total effort of 2,000 person‑hours. Three experimental runs were conducted with effort capacities set to 80 %, 100 %, and 120 % of the baseline. For each capacity, SDO generated 12, 15, and 18 Pareto‑optimal release plans, respectively. These were compared against three baselines: (a) random search, (b) a heuristic based on the value‑to‑effort ratio, and (c) manually crafted plans produced by senior managers. The SDO solutions consistently outperformed the baselines: average satisfaction scores were 15–20 % higher, average dissatisfaction scores were 18–25 % lower, and the structural composition of the plans (mix of high‑value/high‑effort and low‑value/low‑effort features) was more balanced.

A post‑study survey involving 20 stakeholders (product owners, developers, and marketing staff) assessed the perceived usefulness of the SDO output. Participants rated the applicability of the results to actual decision‑making at 4.6 / 5 and the understandability of the trade‑off visualizations at 4.3 / 5. Qualitative feedback highlighted that the explicit presentation of “what we lose if we omit a feature” helped surface hidden risks and align expectations across departments.

The authors discuss several implications. First, incorporating asymmetric evaluation aligns release planning with real stakeholder psychology, potentially reducing post‑release disappointment and rework. Second, the continuous Kano technique offers a practical bridge between qualitative user research and quantitative optimization. Third, the Pareto‑front approach equips managers with a menu of viable options rather than a single “optimal” plan, supporting deliberative decision‑making under uncertainty.

Limitations are acknowledged. The accuracy of (S_i) and (D_i) depends on the quality of the Kano survey; respondents’ understanding of the questionnaire can vary, introducing noise. The current model assumes binary inclusion of features, which does not capture partial implementations or phased roll‑outs. Moreover, the weighting of the two objectives remains a subjective choice; future work could explore automated weight learning (e.g., Bayesian optimization) based on historical release outcomes.

In conclusion, the paper makes a substantive contribution to the field of software release engineering by formalizing and solving the Asymmetric Release Planning problem. The proposed SDO framework demonstrates that simultaneously maximizing satisfaction and minimizing dissatisfaction yields release plans that are quantitatively superior to traditional heuristics and qualitatively valued by practitioners. This work opens avenues for richer stakeholder‑centric planning models and suggests that asymmetric evaluation should become a standard consideration in product roadmap optimization.