AbstrakPabrik Tahu Melati di Batu menghadapi permasalahan inefisiensi produksi akibat ketiadaan sistem yang memadai untuk meramalkan kebutuhan bahan baku kacang kedelai, yang berpotensi menimbulkan kerugian operasional. Kegiatan pengabdian ini bertujuan meningkatkan efisiensi produksi melalui perancangan dan penerapan sistem informasi peramalan berbasis website. Mitra sasaran adalah pemilik dan staf Pabrik Tahu Melati. Metode pelaksanaan mengunakan waterfall meliputi analisis kebutuhan, desain sistem, implementasi pengembangan aplikasi web dan model Double Exponential Smoothing di dalamnya, verifikasi, hingga pemeliharaan dengan pelatihan dan pendampingan. Hasil kegiatan menunjukkan peningkatan signifikan: secara kualitatif, mitra kini mampu mengoptimalkan manajemen persediaan dan membuat keputusan berbasis data. Secara kuantitatif, ketepatan pembelian bahan baku meningkat dari 60% menjadi 95%, frekuensi masalah stok menurun dari 5 kali menjadi 1 kali per bulan, dan staf operasional kini mampu mengoperasikan sistem dengan tingkat pemahaman yang naik dari rata-rata 1,75 menjadi 4,25. Hasil Kuesioner kepuasan mitra menujukkan skor 3,2 dari 4 atau 80% puas dengan pelaksanaan pengabdian ini. Kata kunci: efisiensi produksi; sistem informasi peramalan; double exponential smoothing; pengambilan keputusan berbasis data; UMKM. AbstractThe Melati Tofu Factory in Batu faces production inefficiency due to the absence of a system capable of anticipating soybean raw material requirements, which has the potential to cause operational losses. This community service activity aims to improve production efficiency through the design and implementation of a web-based forecasting information system. The target partners are the owners and staff of the Melati Tofu Factory. The implementation follows the waterfall method, covering requirement analysis, system design, web application development incorporating the Double Exponential Smoothing model, verification, and maintenance, along with training and mentoring. The results of the activity indicate significant improvements: qualitatively, the partners are now able to optimize inventory management and make data-driven decisions. Quantitatively, the accuracy of raw material purchases increased from 60% to 95%, the frequency of stock-related issues decreased from five times to once per month, and operational staff are now able to operate the system, with the average level of understanding increasing from 1.75 to 4.25. The partner satisfaction questionnaire results show an average score of 3.2 out of 4, indicating that 80% of the partners are satisfied with the outcomes of this community service. Keywords: production efficiency; forecasting information system; double exponential smoothing; data-driven decision-making; MSME.