What was known before? • Implementation and use of electronic data (E-data) capture are increasing worldwide. Few published papers have examined the process and learning from large, multi-site observational data collection, especially for facility-based intrapartum care. Design choices may vary according to the purposes, data type, local context, capacity and number of data collectors. What was done? • We applied a five-step framework to evaluate EN-BIRTH study processes including design and use of a custom-built E-data capture system in five hospitals, in three low- and middle-income countries (LMICs), with variable internet connectivity. For this article, we undertook descriptive analyses of relevant study documentation (protocols, operating procedures etc.) and focus group discussions exploring the research team’s experience regarding design and implementation of E-data collection. These findings have implications for E-data development and use in other LMIC settings during research/surveys or programme monitoring. What did we learn from each step? • Step 1) Selection of EN-BIRTH study data collection approach and software E-data capture platforms vary in complexity, adaptability and cost. A systematic selection process is helpful based on purpose, and non-negotiable characteristics in order to achieve the study objectives. EN-BIRTH needed to collect time-stamped clinical observation data for > 23,000 women and newborns in labour wards, operation theatres, and kangaroo mother care wards. Exit-survey interviews were conducted, and register-record and case-note data were extracted. Hence a custom-built system was required since there was no suitable E-data data capture tool available on the market. • Step 2) Design of data collection tools and programming The transition from paper to app-based tools required in-depth consultation with data collectors, various tool users, and piloting, involving an iterative process that took more time than anticipated. Finalising variable lists and permitted data ranges early during software development process were fundamental. • Step 3) Recruitment and training of data collectors Standardised training materials were essential with skills-based sessions focused on the study objectives, research procedures, and competency-based use of the software are key. • Step 4) Data collection, quality assurance, and improvement A collaborative, multi-directional learning network of South-South and also North-South learning was valued and helped by regular, multisite virtual calls, sharing progress by site based on the data monitoring dashboard, and also sharing local solutions with other teams for peer-to-peer learning. Inclusion of facility-level stakeholders in the planning and organisation of data collection was essential to avoid disruptions to routine services. • Step 5) Data management, cleaning and analysis E-data collection was perceived to reduce data cleaning challenges and to reduce erroneous entries however, open text fields and data captured in four different languages requiring back translation were still time consuming during analyses. What next? • Our custom-built E-data tool had advantages including the user-friendly interface, time-stamping, increased data security, real-time monitoring, and inbuilt data quality measures. However, careful assessment of the context and people-time costs are needed and custom-built software should only be considered if existing E-data platforms are not able to meet the objectives of a given research or health programme.