Teguh Arie Sandy, Saeful Anam, Tri Septianto, Angger Binuko Paksi, Muhammad Fadli Prathama, Andi Dahroni, Muh. Nurtanzis Sutoyo, Rahmad Purnama, Heri Nurdiyanto
This study examined Indonesian user sentiment toward Canva by analyzing 415,785 user reviews collected exclusively from the Google Play Store. The objective was to uncover prevailing sentiments, identify common complaints, and explore user expectations regarding Canva's features. The comments were preprocessed and categorized into five main themes: features, security, performance, content, and interface appearance. Overall, users expressed a predominantly positive sentiment, particularly highlighting Canva's ease of use and the variety of design templates. However, recurring negative feedback revealed several concerns, including limitations in the free version, poor performance on low-end devices, data privacy issues, lack of localized content, and interface layout problems. To classify sentiment, two machine learning models were applied using Orange3. The Naive Bayes model performed better, with an AUC of 0.895, an accuracy of 0.902, an F1-score of 0.868, a precision of 0.852, and a recall of 0.902. Logistic Regression yielded lower results, with an AUC of 0.489, an accuracy of 0.903, an F 1 -score of 0.857, a precision of 0.815, and a recall of 0.903. These findings provided useful insights into factors influencing user satisfaction and dissatisfaction, which could inform Canva's future product development to better meet the needs of the Indonesian market. © 2026 IEEE.
Ahlimedia Press, Educational Technology, Malang, Indonesia; Universitas Kiai Abdullah Faqih, Pendidikan Agama Islam, Gresik, Indonesia; Politeknik Negeri Madiun, Computer Science, Madiun, Indonesia; Institut Teknologi PLN, Teknik Informatika, Jakarta, Indonesia; Univ. Sembilanbelas November Kolaka, Sistem Informasi, Sulawesi Tenggara, Indonesia; Universitas Terbuka, Ekonomi Pembangunan, Banten, Indonesia; Universitas Negeri Yogyakarta, Teknik Industri, Yogyakarta, Indonesia
