Sensitivity Analysis of Criteria Weights in Selecting a Backend Web Framework Using the SMART Method

Abstract

There are differing perceptions regarding framework usage. A key challenge in developing enterprise-scale Software as a Service (SaaS) applications is selecting the right backend web framework. This study addresses this challenge by applying the Simple Multi-Attribute Rating Technique (SMART) method. A total of 15 popular frameworks were evaluated based on eight relevant criteria. The criteria used were code quality & Structure, Stability & Mature, Ecosystem & Support, Security, Scalability, Performance, Maintenance Cost, and Ease of Integration. Evaluation data and initial scores (scale 1–10) were collected using expert judgment by five experts through a standardized utility value function, synthesized with quantitative testing data from TechEmpower Web Framework Benchmarks and empirical metrics from GitHub repositories. The initial SMART scores placed Spring Boot (Java) and ASP.NET Core (C#) as the top choices, followed by Django (Python), Laravel (PHP), and NestJS (Node.js). To validate these decisions, a Sensitivity Analysis was conducted by varying the weighting of the security and performance criteria. The combination of the SMART method and sensitivity analysis offers a transparent and adaptive framework for multi-criteria decision-making. The first-ranked framework, Spring Boot (Java), demonstrated a very high level of robustness to changes in evaluation criteria priorities. ABSTRAK (Bahasa Indonesia) Terdapat perbedaan persepsi mengenai penggunaan framework. Permasalahan kunci dalam pengembangan aplikasi Software as a Service (SaaS) berskala enterprise adalah memilih framework web backend yang tepat. Studi ini mengatasi tantangan tersebut dengan menerapkan metode Simple Multi-Attribute Rating Technique (SMART). Sebanyak 15 framework populer dievaluasi berdasarkan delapan kriteria relevan. Kriteria yang digunakan adalah kualitas kode & Struktur, Stabilitas & Mature, Ekosistem & Dukungan, Keamanan, Skalabilitas, Performa, Biaya Maintenance, dan Kemudahan Integrasi. Data evaluasi dan skor awal (skala 1–10) dihimpun menggunakan metode expert judgment oleh 5 orang pakar melalui fungsi nilai utilitas terstandardisasi, yang disintesis dengan data pengujian kuantitatif TechEmpower Web Framework Benchmarks serta metrik empiris repositori GitHub. Hasil skor SMART awal menempatkan Spring Boot (Java) dan ASP.NET Core (C#) sebagai pilihan utama, diikuti oleh Django (Python), Laravel (PHP), dan NestJS (Node.js). Untuk memvalidasi keputusan ini, Analisis Sensitivitas dilakukan dengan memvariasikan bobot kriteria keamanan dan performa. Kombinasi metode SMART dan Analisis Sensitivitas menawarkan kerangka kerja transparan dan adaptif untuk pengambilan keputusan multi-kriteria. framework peringkat pertama yaitu Spring Boot (Java) menunjukkan tingkat kekokohan (robustness) yang sangat tinggi terhadap perubahan prioritas kriteria evaluasi.