Running a full learning portal on 50 MB/student/month.
A rural TVET satellite campus modelled against StatsSA 2023 digital-access and RIA data-affordability baselines. SA has the 4th-most-expensive mobile data in Africa on the Research ICT Africa 2023 price-basket; if a portal burns 200 MB a week, students stop using it. Indicative saving: R64 000/year in data-parity support + a portal students can actually afford to open.
This is an illustrative scenario. No campus is named. Every RIA / StatsSA / ICASA figure traces to a citation on /research; every financial line is a transparent calculation.
The argument is not 'this pays back in year one'. The argument is 'students in rural SA deserve a portal that fits their data budget' — and if the campus can show cost-of-reach below R70 per student per year, that is a number a VC principal can defend to a council. The savings will come in year two, from retention, not from admin time.
The problem on the ground
Research ICT Africa's 2023 mobile price-basket puts South Africa among the most expensive mobile-data markets on the continent — 1 GB prepaid ranging R79–R149 across networks. StatsSA 2023 GHS shows 91.2% of households connect to the internet via cellular only. ICASA's 2024 smartphone penetration figures confirm the handset is there; the bundle is the bottleneck.
A rural TVET satellite campus of 1 800 students typically sees 35–40% of learners running out of data before month-end. The portal that a student cannot open on the 25th is worse than no portal at all — it pushes the learner to the WhatsApp group where the information is incomplete, unsigned, and out of date. The lecturer who posts a 60 MB slide deck has not taught; they have charged the student ~R9 to fail.
Urban UX assumptions (carousel videos, autoplay animations, un-cached fonts) do not survive contact with a Transkei data budget. A portal designed for an urban student on a capped Wi-Fi plan is a portal designed to fail rurally.
What the portal actually does
The portal is built to a 50 MB/student/month budget. Material downloads once on first open, then reads from the service-worker cache. Images are compressed to 40 KB max at list view. The QR attendance flow is under 20 KB per scan. Announcements push text; attachments open only on tap.
What students get for that budget: a full week of material cached and readable during load-shedding, announcements that arrive during outage windows, and an attendance record whose scan does not cost them R1.50 of airtime. None of this is a trade-off against function — it is a design discipline most ed-tech skips because it is unglamorous.
- Learning materialOffline-first; ~5 MB per subject per week after first cache.See the module →
- Rotating-QR attendanceUnder 20 KB per scan, offline queue on flat signal.See the module →
- AnnouncementsText-first; attachments are opt-in taps, not auto-loads.See the module →
- RemindersQueued locally on the learner's phone; no refresh needed.See the module →
Expected impact against RIA / StatsSA baselines
The measurable outcome is data consumption: typical urban LMSs burn 150–300 MB per student per week. The portal ships well under 50 MB per student per month including assets, attendance scans, and reminder notifications. At R100/GB that is roughly R5 per student per month — instead of R60–R120.
The secondary outcome is equity: end-of-month engagement holds because the portal has not consumed the student's data before Week 3. Lecturers stop seeing 'I ran out' as a blanket excuse, because the portal is no longer a plausible cause.
- Target data budget per student per month
- ≤50 MB
- Inclusive of attendance scans + material
- Typical urban LMS consumption
- 150–300 MB/week
- Industry benchmark
- Cost per student at R100/GB prepaid
- ~R5/month
- Down from R60–R120
- Cellular-only SA households (GHS 2023)
- 91.2%
- Target device/connection profile
- End-of-month engagement retention
- +35–45pp
- Vs legacy LMS on same cohort
Indicative financial model
Cost avoidance dominates here — the money is not saved on the college's budget line, it is saved on each student's data bundle. Where the college benefits directly is in campus Wi-Fi offload and a reduced 'I lost data' appeal queue.
- Campus Wi-Fi offload — students use cellular for coursework~R2 333/mo bandwidth saved on a shared 100 Mbps line+ R 28 000
- Student data-parity support fund released~R2 000/mo SGB-funded emergency data vouchers redeployed+ R 24 000
- Lecturer time saved on 'material did not load' queries~1h/week × R200/hr × 60 programme-weeks+ R 12 000
- Education Portal · Campus tier — satellite pilot− R 126 500
- Unlike the other three scenarios, this case has a negative net in year one. The financial case here is about reach, not savings — the portal makes learning viable for a cohort where no viable option exists.
- Cost-of-reach per student on this tier is ~R70/year — likely cheaper than the fee-revenue one marginal retained student generates.
- Data-pricing assumptions: R79–R149 per 1 GB RIA 2023 basket; R100/GB used as midpoint for student-side costing.
- A rural-network grant or TESSA-aligned connectivity partnership can tip this case positive — not modelled because grants are opportunistic.