Remote sensing observations such as normalized difference vegetation index (NDVI) trends can provide important insights into past and present land condition. However, they do not directly provide comprehensive information about our representation of land degradation and the processes at work. This study aimed to analyze vegetation productivity underlying factors in order to assess land degradation and to highlight the impact of definitions on its quantitative assessment, using Mozambique as case-study. Land productivity change were first analyzed using NDVI time-series (2000-2016), and a two-step framework was then used to understand the main factors of these productivity changes. The impact of land degradation's definition was assessed based on four types of stakeholder, with different priorities in terms of ecosystem services. The results show that 25% of the country display a significant land productivity decrease, while only 3% display a land productivity increase. A large part of these land productivity changes (>61% of the decrease, and >98% of the increase) is directly assigned to human activities, such as native forest growth or tree plantations (for the increase), or forest degradation, deforestation and loss of grassland productivity (for the decrease). We showed that the fraction of degraded land varies according to stakeholders' definitions, ranging from 12% to 20% of the Country, much less than the 39% estimated by Tier 1 United Nations Convention to Combat Desertification. This study provides a sound methodological framework for assessing land degradation status that could help stakeholders to design national and locally relevant land degradation mitigation policies or programmes.