The use of SMS messaging for education has grown in recent years, with particular attention recently during the Covid-19 pandemic. Mobile phones often have high levels of ownership in low-income contexts compared to computers, and lower connectivity requirements, which arguably make this a more equitable medium than data-heavy online instruction, for example. However, given that gender can be a factor to influence mobile device access and use, it is also important to consider educational applications through a gender lens, to avoid further exacerbating digital divides. In this paper, we present an analysis of server log and evaluation data in relation to a literacy-focused initiative for primary-aged learners carried out in Kenya as part of the Tusome programme and through the SMS-based M-Shule education platform, which does not require an Internet connection or smartphone to run. The extent of engagement with the platform varies according to gender and location within the country. The data also demonstrate a positive impact on learning outcomes regardless of learners' gender and location. Furthermore, the learning gains are shown to be relatively cost-effective in comparison with educational technology interventions in similar contexts. The findings show that this low-connectivity adaptive model has a positive impact on learning outcomes. It is a scalable approach to support a range of learners in Kenya, providing more support to learners who need it, and leading to increased foundational learning outcomes overall. As such, the findings will also be of highly relevant to other low-connectivity contexts.