I have sat in many programme review meetings to know how this story mostly ends.

A development programme partners with an agtech. Farmers are onboarded. Advisory messages are sent. App downloads climb. Field staff visit villages. The logframe target was 40,000 farmers- we reached 50,000. Programme successful.

Then the programme ends.

Six months later nobody goes back to check. But if they did, they would find that most of those 50,000 farmers have quietly stopped using the platform. They are back to calling the input dealer. Back to asking the neighbor. Back to doing what they always did before the programme arrived.

The agtech still has the 50,000 number in their pitch deck. The donor has filed a satisfactory completion report. And the farmer for whom the solution was designed has moved on.

This is a donor-funded captive engagement. Generally, you serve the ones who fund you, so in this case, the farmer was never a customer. The farmer was a programme target who showed up because the field worker asked, because there was a free input attached, because it cost nothing and might lead somewhere.

The moment it costs something the farmer makes a decision about whether it is worth it. Most of the time the answer is no. The value was never proven on their specific plot, in their specific season, for their specific crop.

The path of least resistance

I want to acknowledge the genuine barriers here. Agricultural income is seasonal. Ticket sizes are very small. Noone is arguing the potential of agtech solutions. However, the value of most agtech solutions is still probabilistic- better yield may be, lower cost may be- with no guarantee. There is also a first-mover disadvantage- the first agtech to charge in a market where everything has always been free faces enormous resistance regardless of how good their solution is. These are real barriers. The path is longer and more expensive.

So, most of the agtech takes the programme. And then another programme. And then another. And somewhere along the way the business model becomes find the next funder before the current one runs out.

Three models

Let me be clear about something before I go further.

I think there is nothing inherently wrong with building an agtech on programme money and selling it to the highest bidder. Build a large farmer base, generate proprietary data, develop a distribution network, and exit to an input company or a larger platform. That is a legitimate venture capital play.

There is also nothing wrong with accepting that Indian smallholder agriculture will remain programme-dependent for the foreseeable future. Building a sustainable business within that ecosystem by winning programmes, delivering well, building institutional credibility is also fair.

And there is the farmer-pays model. The model that scales without a ceiling and genuinely proves that what you built is worth something to the person it was built for.

All three models are fair. All three require different strategies and different measures of success.

The problem is pretending we are running one model while actually running another. Pitching to impact investors as a farmer-centric business while optimising for programme metrics and exit. Telling donors we are building sustainable farmer engagement while knowing the engagement will disappear the moment the programme ends.

The scale illusion

We talk about scale constantly. Reaching a million farmers. Ten million farmers. Transforming Indian agriculture.

If the revenue model depends on donors and programmes, scale will be limited by how much donor money exists. And donor money is finite, cyclical, and increasingly competitive.

Some genuinely valuable businesses scale through models where the end user does not pay directly- Google is the most obvious example. The question is whether the value created is real and whether the business model that captures that value is sustainable. A business that charges input companies for access to verified farmer data while providing genuinely useful agronomic advice to farmers for free could be both sustainable and beneficial to farmer. What I am arguing is that most donor-funded agtech engagement does not look like this. It generates programme metrics and programme metrics disappear when the programme ends.

Why farmers do not pay

It is easy to blame the farmer. Low digital literacy. Resistance to change. Short term thinking. That framing, in my opinion, is wrong and lazy.

Farmers make extremely sophisticated risk calculations every season under conditions of deep uncertainty. They are experienced. Their experience tells them to be sceptical of solutions that arrive with a programme and leave when the funding runs out.

They have seen this before. Many times.

The real reason farmers do not pay is that nobody has yet shown them something worth paying for in a way that matches how they earn and measure value.

If a solution genuinely increased a farmer’s income by Rs. 10,000 per season, the farmer would find a way to pay Rs. 1,000 for it. That is basic economics. The fact that farmers resist paying is diagnostic and it tells us the value has not been proven convincingly enough on their terms.

What actually works

The cases where farmer engagement survives programme end have a few things in common.

The solution solved a problem the farmer already knew they had. Price discovery. Faster credit when the input dealer runs out of stock. Transparent weighing at the mandi. When you solve a felt need the farmer does not need to be convinced. They come looking for you.

The farmer paid something from day one. Even Rs. 50 or Rs. 100. A token payment creates a completely different relationship than free. When we pay for something, we use it differently. We hold the provider accountable. We notice when it does not deliver. The act of paying transforms the farmer from a beneficiary into a customer and customers demand. They push back, provide feedback and that feedback makes products better.

The agtech had skin in the outcome. Outcome based pricing changes everything. If the recommendation works, the agtech earns. If it does not neither does the agtech. That alignment of incentives is the an honest version of partnership. Rest is service delivery with good intentions.

The institution owned the relationship. When a farmer producer organizations or cluster level federations collectively subscribe to a service, holds the agtech accountable, and integrates the tool into its regular operations- the engagement survives programme end because the institution survives. Looking at India’s smallholder farmers, investing in strong farmer institutions is a prerequisite for agtech sustainability.

Path forward

Every programme that funds agtech engagement should ask one question at design stage: what happens to this engagement on day one after we leave?

If the honest answer is that it stops, the programme should be designed completely differently.

What does that look like in practice:

• Build the farmer-pays model in from the beginning, even if the programme subsidizes it initially. The subsidy should be explicitly time-bound and structured to reduce over time creating a path toward sustainability rather than a permanent dependency. Experiment with payment mechanisms that match agricultural realities such as seasonal payment cycles tied to harvest, FPO-level subscriptions that aggregate individual farmer contributions, outcome-based fees that only trigger when the farmer’s income demonstrably improves.
• Measure retention after programme end, not just reach during it. If the M&E framework does not include a 6 to 12-month post-programme follow-up on farmer engagement- we are measuring the wrong thing.
• Invest in building strong institutions and platforms such as farmer producer organizations, cooperatives, cluster level federations that outlast the funding. Collective institutional behavior is durable. The engagement can survive when the institution survives.

None of these are easy but I think all of them are more honest than pretending the problem does not exist.

The agtech sector has a business model problem that has been papered over by a decade of development funding that asked the wrong questions and rewarded the wrong metrics.

Charging farmers is hard. I am not pretending otherwise. The path is longer and more expensive than most business plans acknowledge. Avoiding hard problems defers them until the programme ends, the funder moves on, and the farmer goes back to calling the input dealer.

The farmer has seen enough projects arrive and disappear. What they need is something worth paying for.

Disclaimer: The ideas and observations in this piece are entirely my own. I used AI to help structure them.