Field evidence on your own farmers and plots
We put your questions to your farmers, check each plot against the satellite and weather record, and hand back a report with the row-level data attached. Every figure names the product that produced it and the limit it holds to. Certifiers, EUDR exporters, lenders and programme funders read this work; what we hand them gets checked.
The 3 steps
Only the first one needs anything from you.
- You send the listFarmers with phone numbers, plots with polygons, or both. Your questions arrive as a spreadsheet. The list stays yours; processing is under a written agreement.
- We gather the evidenceInterviews over WhatsApp or through a link your officers open with the farmer. Every plot is screened against public satellite and weather products at its own coordinates.
- You get the report and the recordOne page saying what needs action, the flagged rows with their evidence, and one row per farmer and per plot as CSV and GeoJSON.
The 6 services
Priced per unit of evidence — an interview, a plot, a study — and agreed before anything starts. Every service on this page is quoted on request. All of them run in Kenya, Tanzania, Uganda and Rwanda, with one exception: respondent incentives can only be paid in Kenya.
Quotes instead of a rate card
Datasets and analyses are a machine answering a key: they carry a rate card, and you serve yourself the moment you hold one. A validation puts people and field time against your own list, so it is scoped before it is priced. The figure moves with the number of farmers, the kind of questions, the smallest plot you will accept and who the report is for. That is why this page carries no number.
Work still in development
These are listed apart from the 6 services: none of it is available yet, and we do not quote for work that has not run.
| What | Where it stands | Service |
|---|---|---|
| Voice interviews | The same questionnaire conducted as a spoken conversation, for farmers who do not use WhatsApp. The interview engine already treats the channel as swappable, so a call will arrive on the same path as everything else. It is not available, and we do not quote for it until it has run. | Farmer interviews in East Africa |
| Forest against tree crop | Separating a coffee replanting or a eucalyptus harvest from the clearing of real forest. No public forest-change product distinguishes them. We hold a trained classifier for perennial crops and are wiring it in; until that is done, every detection carries cause-unverified and a test in the code asserts that no causal verdict can be emitted by accident. | Plot verification and EUDR forest checks |
| Radar alerts as a second opinion | Near-real-time radar detections see through cloud and catch smaller clearings than the annual product. The module is written and dormant. These alerts exist only inside primary humid forest, so over ordinary farmland they add nothing, and a radar zero must never be read as a clean result. | Plot verification and EUDR forest checks |
Claims we do not make
A buyer who has seen other offers tends to ask about these next. If an offer includes one on the strength of satellite imagery alone, ask how it was checked.
- That a spray, a fertilizer application or an irrigation event happened, or how much was applied. No satellite sees these at any resolution or any price.
- A per-farmer yield in kilograms as a measurement. At smallholder field sizes the error is most of the quantity itself.
- A crop identified from satellite beyond tea against not-tea and perennial against annual.
- A cause for any tree-cover loss. That is settled by a visit.
- That a plot has been cropland continuously. The cropland baseline covers 2019 and 2020 and nothing since.
- A planting or harvest date read from imagery. Those dates come from the farmer, and imagery offers a window at best.
- A confidence score on any answer. Nothing here has been calibrated against a labelled set, so the number would be invented.
Scoping a study
A validation starts with a short conversation about what you need to show, to whom, and by when. From that we can say which of the 6 services the question needs, and which it does not, before anything is priced.
You remain the controller of any list you share. We process it under a written agreement signed with you, separate from our own registration with the Office of the Data Protection Commissioner in Kenya. It is never re-used for anything else. The privacy policy sets out the rest.