Challenge
One major barrier to collecting and using GDD is existing data structures. What should you do if existing accounts were configured without a gender identifier? Certainly, one could add a gender identifier to every new account, but this would not allow you to analyze established users. You could also ask existing users to identify their gender, but this is impractical (particularly for dormant accounts) and introduces friction in the user experience. Even where gender information was collected during account sign-up but not associated with an account on the backend, manually associating this information would prove time-consuming and costly.
Information on mobile users by gender is limited because of social norms and incomplete data. This presents challenges for MNOs as they seek to improve their service offerings for women.
Approach
The GSMA Gender Analysis and Identification Toolkit (GAIT), developed in cooperation with Dalberg Data Insights, aims to use machine learning to identify the gender of mobile users. [1] This allows mobile operators to better understand their user base, tailor products to their user base, and, as a result, help close the mobile gender gap.
Results
Using software informed by a survey of users to identify “ground truth” gender data, the GAIT can predict MNO user gender with greater than 80% accuracy. GSMA piloted the tool with Robi, an MNO with 40% market share in Bangladesh. GAIT predicted user gender with 84.5% accuracy. In the “ground truth” survey process, Robi found that 78% of female users had been incorrectly labeled as men in the sign-up process, further underscoring the value of the AI tool. GDFSPs and IPSs could employ a similar approach to analyze user behavior and estimate the gender of existing users.
[1] “The GSMA’s Gender Analysis and Identification Toolkit (GAIT),” GSMA, August 31, 2025, https://www.gsma.com/solutions-and-impact/connectivity-for-good/mobile-for-development/gsma_resources/the-gsmas-gender-analysis-and-identification-toolkit-gait/.