This summer we shipped the biggest FarmerChat update yet that we lovingly call “V2”. V2 is now live for farmers in every country we serve, and it’s already changing how they use it. Here’s an honest look at the first few weeks, and what the early numbers are starting to tell us.
V2 is a ground-up rebuild of the assistant. It’s faster, the conversations feel more natural, and it’s far better at understanding what a farmer is actually asking, whether they type, talk, or send a photo. We rolled it out market by market and watched closely. The short version of what we found: we didn’t suddenly reach more farmers. The farmers already using FarmerChat got much more out of it.
More farmers are actually using it
The first thing we look at is activation: of the people who open FarmerChat, how many go on to ask a real question? It sounds basic, but it’s where most apps quietly lose people. Roughly one in five people who download an app never even open it, and plenty who open it never do anything at all. With V2, activation climbed in every major market we measured, from around half of users to about two-thirds. Farmers aren’t just landing on the app anymore, they’re starting a conversation.
By the numbers
| Market | V1 (before) | V2 (now) | Lift |
|---|---|---|---|
| India | 36.6% | 58.9% | +61% |
| Kenya | 37.7% | 55.8% | +48% |
| Ethiopia | 45.1% | 60.2% | +33% |
| Nigeria | 44.8% | 57.4% | +28% |
And they’re asking more
The farmers who engage aren’t stopping at one question. Across markets, people now ask about one and a half times as many questions as they did on V1. That matters more than it might seem: a farmer who asks a second and third question is a farmer who’s finding the answers useful enough to keep going. A single question can be curiosity. A back-and-forth is trust. Depth, not just downloads, is the signal we trust most. And that deeper use matters: farmers who engage more with FarmerChat are also more likely to put the practices they learn about into action.
Ethiopia leads the way
Ethiopia is the clearest example. Farmers there now ask around four questions per visit, up from about two and a half, the biggest jump of any market. We don’t think that’s luck. It’s what happens when better language support meets a conversation that finally feels like it’s listening, and it’s the kind of result we want to reproduce everywhere we work.
| Market | V1 (before) | V2 (now) | Lift |
|---|---|---|---|
| India | 1.3 | 2.6 | +100% |
| Kenya | 1.4 | 2.2 | +57% |
| Ethiopia | 1.8 | 3.5 | +94% |
| Nigeria | 1.4 | 2.1 | +50% |
| Cohort | V1 (before) | V2 (now) | Lift |
|---|---|---|---|
| Week 10 | 8.7% | 14.8% | +70% |
They keep coming back
Getting a farmer to ask a question is one thing. Getting them to come back weeks later is the real test, and it’s the number we care about most, because advice only compounds if people return for it across a season. So we followed our longest-running group of farmers for ten weeks and compared V2 against the version before it.
The gap grows over time. Both versions start in almost the same place, then steadily pull apart, with V2 holding on to far more of its most engaged farmers week after week. By week ten, more than one in five of our power users were still active every single week, roughly double the previous version. Again, the headline isn’t that we reached more people. It’s that the people we reach are staying.
Built to feel human
Some of the biggest improvements are the least technical. FarmerChat now asks a clarifying question or two before it answers, the same way a trusted shopkeeper asks your acre size or asks to see a photo before giving advice. In testing, farmers found that completely natural, because it’s how good advice has always worked. And when we sat with them, voice was the first thing they reached for, not text. As one farmer put it, voice is king. So we’re shaping the whole experience around how farmers actually talk, not how software usually expects them to.
What’s next
V2 is the foundation, not the finish line. The next step is teaching FarmerChat to do more than advise, to help a farmer act on what they learn, and to keep closing the small gaps that still trip people up, like making voice work flawlessly every time. It’s still early, and we’re holding ourselves to proving these gains hold as more farmers come on board and as the seasons turn. But the direction feels right, and the first weeks are the most promising we’ve had.
For now, FarmerChat reaches more than 1.7 million farmers across six countries in sixteen languages, and around seventy percent of them act on the advice they receive.
Curious? Meet FarmerChat and see how it works for yourself. Or visit farmer.chat to try it directly.
