A full performance assessment of 3D‐mapping–aided (3DMA) GNSS in dense urban areas is presented. This first part of a two‐part paper focuses on the effects of algorithm design and user equipment, based on data collected in London using Leica Viva GS15 and u‐blox EVK‐M8T GNSS receivers and a Nexus 9 Android tablet. Best performance is obtained by combining shadow matching with likelihood‐based 3DMA GNSS ranging using hypothesis‐domain integration. Improved versions of the algorithms, together with a comprehensive tuning process, are described. Root mean square horizontal position errors obtained using data from Leica, u‐blox, and Nexus receivers are 3.5, 4.7, and 4.9 m, respectively, compared with 23.6, 26.4, and 31.0 m using conventional GNSS positioning, about a factor of six improvement. Optimal algorithm tuning depends on the environment and the impact of varying grid spacing of the candidate positions is assessed. Algorithms have also been shown to operate in real time on both a Raspberry Pi 3 and a Samsung Galaxy S8+ Android smartphone.