Urban informality, which is prevalent in Africa's rapidly growing cities, can reduce private investments, lower tax bases, and exacerbate urban disamenities. A key policy tool to address this problem is greenfield urban planning, where governments purchase cheap agricultural land on the urban fringe and partition it into planned, surveyed, and titled de novo plots, which people can purchase and build houses on. Yet, there is very little systematic evidence on the effects of de novo planning choices, such as the size and configuration of residential and non-residential plots. This paper studies the consequences of the planned layout of Tanzania's "20,000 plot" project, which provided over 36,000 residential plots in 12 project areas on the fringes of Dar es Salaam in the early 2000s. To study this project, we use new data from questionnaires and satellite imagery from circa 2020 and combine within-neighborhood analysis and spatial regression discontinuity designs. We find that small plots, which command higher land values and are built more intensively, are under-provided; smaller plot owners value homogeneity in plot size, grid structures of layout of blocks are valued, as is access to major paved roads; and public service provision lags the plans. About half the plots are still unbuilt; yet the areas nevertheless attract highly educated owners. These findings suggest that while the project led to large overall gains in land value, significant improvements to planning may be possible.