Well-evaluated digital agriculture programmes in Africa tend to show that farmers can be informed cheaply, and that being informed seldom makes anyone pay again. That is the gap between a pilot and a procurement line: someone must keep paying after the grant ends, and the evaluations rarely find a farmer willing or able to be that someone.
Key takeaways
- Information alone moved behaviour in some trials and not in others. In western Kenya, SMS and field days did not raise fertiliser purchases, while a sugarcane SMS trial raised yields, only for the effect to fade.
- The programmes that persist have a payer with a reason to care: a state that runs a subsidy, a buyer that needs volume, or an input company that sells through the channel.
- Evidence is mixed and often unpublished. Preliminary reports on the Rwanda cooperative trial conflict.
- Ask who pays in year three and for what, before judging the technology.
What the evaluations found
Western Kenya, information trial (2014 to 2017). Researchers Fabregas, Kremer, Robinson and Schilbach tested ways of delivering recommended-fertiliser advice in Kakamega county. A 3ie evidence summary says field days raised awareness and changed beliefs about how profitable the recommended fertiliser was, but purchases did not rise. SMS did not raise knowledge or use of recommended inputs. It gives no cost figures. It does note the extension body, KALRO, reviewed its messages and field days afterwards and partnered with a telecom operator.
Sugarcane SMS and its limits. J-PAL's policy insight, last updated September 2023, reports that SMS to contracted sugarcane farmers raised yields by about 8% at roughly US$0.02 per text and about US$54 more revenue, but that the effect was not sustained in the second season. It also reports that in India SMS about a flood-tolerant rice variety did not improve adoption, and that price information alone did not change the prices farmers got. Cheap per message, yes. Recurring value to the farmer, not demonstrated.
One Acre Fund in Kenya (2021 long rains). A registered trial across 16,170 potential clients at four shops sent SMS reminders with a coupon for an early discounted input bundle. Registry data show shop engagement of 16.4% and 11.5% in the two treatment arms against 12.2% for control, while administrative bundle uptake was about 0.1%. Farmers who also got an in-person discount offer bought more bundles. Its authors conclude digital-only is more cost-effective, but that in-person contact is needed to close the adoption gap. This registry entry is a summary, not a peer-reviewed paper.
A pooled view. A 2025 meta-analysis in Frontiers covers 18 studies, 10,757 farmers and 12 countries, seven of them African. It finds a small effect on adoption from standalone mobile extension (about 2%) and a larger one when combined with traditional extension (about 17%), with very high heterogeneity (I-squared of 93%). The authors flag a small evidence base, and the odds-ratio-to-percent framing on the page is unclear, so read it as direction, not size: mobile works better as an add-on to human contact than as a replacement.
Who pays when it does persist
Rwanda offers the cleanest case study of a payer. The Smart Nkunganire System (SNS) is a public-private partnership between BK TechHouse, a Bank of Kigali subsidiary, and the Rwanda Agriculture Board (RAB), developed in 2017. KT Press reported in March 2021 that more than 1.2 million farmers (a later CGIAR report, from November 2025, cites a higher registered number) and 1,300 dealers had enrolled over three years, that BK TechHouse is the sole provider, and that a Gates Foundation grant was funding a 2021 to 2023 scale-up. Because the system sits inside a government input subsidy, the state is the recurring buyer and the farmer is a user. A 2024 CGIAR user-experience report from the International Potato Center looks at how farmers, agronomists and agro-dealers experience it. We read it only at the abstract level.
Nigeria's earlier e-wallet, the Growth Enhancement Support Scheme, shows the other side. An IFPRI working paper by Olomola found its first-year (2012) achievements below expectations; we read only the abstract. A government payer does not guarantee delivery.
| Payer | Why it pays | Evidence in this article | Weak point | |---|---|---|---| | Government subsidy | Input access, political reach | Rwanda SNS; Nigeria GESS | Delivery quality, budget cycles | | Off-taker or buyer | Needs volume and quality | Sugarcane SMS to contracted farmers; Rwanda cooperative trial | Only farmers inside the contract | | Input supplier | Sells more product | One Acre Fund bundle trial | Digital reminders had low uptake alone | | Donor or DFI | Programme goals | Gates grant for SNS scale-up; IFAD-supported skills work | Ends with the grant | | Farmer | Direct value | Kenya information trial (no purchase effect) | Not demonstrated in the trials above |
AGRA's 2024 status report launch says the private sector drives about 80% of Africa's food economy and that MSMEs manage about 85% of value chains. It lists M-Kulima, M-Pesa, Hello Tractor and AgroCenta as digital tools in use. That tells you where money moves. It does not tell you which tools get paid for repeatedly.
The Rwanda cooperative trial: a warning about conflicting evidence
Innovations for Poverty Action lists a randomised trial with the World Food Programme and the Farm to Market Alliance across 180 cooperatives, 90 treated and 90 comparison, running 2020 to 2024. It shows no results. A November 2024 seminar abstract from Syracuse University says the programme "immediately increased the probability of selling into the formal output value chain" and "dramatically increased revenue by 150%-300%, driven largely by large increases in land under cultivation and input usage". We also saw a report that an earlier brief found no sales or price impact, but we could not open it, so we do not rely on it. A researcher's abstract is not a published paper. Until the final results appear, treat this as a promising and unresolved case.
A separate pilot used SNS records in 40 cooperatives to test whether higher output prices raise input use, per the ATAI project page (Robinson, Spearot, Aggarwal, Godlonton; June 2019 to November 2021). It lists no results either.
Our reading
Our reading: bad technology is rarely the common failure. pilots stall when the buyer in the pilot is a project, and the user who should become the buyer sees no change in what they earn. Where a payer's own revenue depends on the service, as with a state running a subsidy or a buyer sourcing from cooperatives, programmes survive. Where the farmer is expected to pay for information, the evidence above is thin.
Against this, the strongest counterargument is that these are mostly free or subsidised services tested in short windows, so low payment says little about what a farmer would pay for a service that reliably raises income. That is fair. Yet the sugarcane result shows an effect existed. It also faded, and nobody in these studies shows a farmer renewing it at full price.
A caution on scale-up from capacity-building: the Development Gateway reflection on an IFAD-supported Digital Agriculture Skills programme reports 19 training sessions with 940 participants across 15 countries, and lessons that under-resourced project teams struggled and ownership mattered for sustainability. Training users does not create a budget line.
Buyer guidance
If you are a buyer (input company, off-taker, ministry, DFI programme officer):
- Ask who funds the service in year three and write the answer into the contract. If the answer is "a new grant", you are buying a pilot.
- Ask for behaviour outcomes (inputs bought, volume sold, yield recorded), not downloads or messages sent. The Kenya trial shows awareness can rise while purchases do not.
- Check the comparison group. A before-and-after number without one is a story, not evidence.
- Ask what happens when the farmer does not respond to a digital nudge. One Acre Fund data suggest a human channel may be required.
- Check data rights. In subsidy systems such as SNS the platform holds farmer and dealer records; know who owns them if the contract ends.
If you are a founder:
- Find the party whose revenue rises when your user does better, and sell to them.
- Show cost per behaviour change, not cost per message.
- Publish what failed. Buyers have read enough glowing pilot reports.
For shortlisting, the agritech category and the directory list companies by focus and country; use Find Partner to describe the buyer you are and request references that name a paying customer outside a grant.
Frequently asked questions
Do farmer SMS services work? Sometimes. A sugarcane trial raised yields by about 8% at low cost, but the effect did not last into the second season, and a Kenyan fertiliser-advice trial found no effect on knowledge or purchases.
Who pays for agritech tools in Africa? In the evidence reviewed, mainly governments running subsidy schemes, buyers that need volume, input suppliers and donors. Farmers paying directly is rare in these studies.
Is the Rwanda cooperative trial proof that integrated programmes work? Not yet. Preliminary claims are from a seminar abstract, an earlier brief reportedly found no sales or price impact, and the final results are unpublished.
What should a buyer ask a vendor first? Who pays for the service in year three, and what behaviour outcome has been measured against a comparison group.
Sources
- 3ie - Improving how farmers receive agricultural information in Kenya (institutional)
- J-PAL - Agricultural information and extension services (institutional)
- Precision Development (PxD) registry - One Acre Fund Kenya early discounted bundle trial (technical)
- Innovations for Poverty Action - Connecting smallholder farmers to agricultural value chains in Rwanda (institutional)
- Syracuse University Maxwell School - Spearot seminar abstract (technical)
- KT Press - BK TechHouse: new digital agriculture roadmap launched (media)
- ATAI - Rwanda Smart Subsidy System experiment (technical)
- CGIAR / International Potato Center via CGSpace - Smart Nkunganire System user experience (technical)
- IFPRI via RePEc - Nigeria's Growth Enhancement Support Scheme (institutional)
- Frontiers in Sustainable Food Systems - Mobile phone-based extension meta-analysis (technical)
- AGRA - Africa Agriculture Status Report 2024 launch (institutional)
- Development Gateway - Reflections on the IFAD Digital Agriculture Skills programme (institutional)