The rules first, and the marketplace second.
Every agricultural data business that has gone wrong went wrong in the same way: it collected first and wrote the rules afterwards, at which point the rules were written to permit whatever had already been collected. These are written first, which is the only order that means anything.
The line, drawn before anything is collected.
What could be sold
- Aggregated, anonymised yield and soil benchmarks, at regional grain
- Variety performance by soil type, across many farms
- Anonymised drilling-date and blackgrass-pressure analyses
- A stated revenue share back to every farm in the aggregate
- A published anonymisation policy, with the grain size in it
- Deletion on request, including from future aggregates
What could never be
- Any identifiable single farm’s data, to anybody, ever
- Field boundaries tied to a named business
- Anything to an insurer that could reprice an individual farm
- Anything to a landlord about a tenant’s ground
- Anything to an input supplier that could be used to price to an individual
- Anything at all without a signed, revocable, per-season consent
The two entries that matter most.
Nothing to an insurer that could reprice an individual farm, and nothing to a landlord about a tenant’s ground.
Both are the same failure: data collected for one purpose, used to alter the terms on which somebody is allowed to farm. A yield map showing five poor years is a perfectly good agronomic record and a devastating document in a rent review.
Any marketplace that does not have those two lines in it will eventually sell into both markets, because they are where the money is. Writing them down first is the only defence, and even then it only works if the rules are published where the people affected can read them.
What happens when somebody changes their mind.
| If a farm | Then |
|---|---|
| Withdraws consent | Their data leaves the pool within thirty days, and every future aggregate is recalculated without it |
| Asks what has been sold | A complete list, per dataset, per buyer, per date. Within a week |
| Asks for deletion | Deleted, including from backups on the ordinary backup cycle, with the date confirmed |
| Objects to a specific buyer | Excluded from that buyer’s aggregates specifically, which is fiddlier and is the point |
| Sells the farm | Consent does not transfer. The new occupier consents afresh or the data stops |
| Dies | The data stops immediately, pending whatever the estate decides. This is not a hypothetical for a farming population with an average age near sixty |
Why a farm is the wrong party to run this.
Because a farm that aggregates its neighbours’ data has a conflict of interest that no policy document resolves. It competes with them for land, for contracts and occasionally for the same buyer.
The right structure is a farmer-owned co-operative or a genuinely independent body, with the farms holding the governance rather than a brand holding it on their behalf. That is slower, less profitable and considerably more likely to still exist in ten years.
Which is the conclusion the soil-data page reaches by a different route, and it is why neither of these two pages ends with a checkout.
None of this exists in this build
No data is collected, aggregated, sold or shared anywhere on this site. This build takes no payment, holds no member records, and the basket is forgotten when the tab closes. These pages describe what the rules would have to be, not what any system is doing.
The dataset itself.
Six years of yield maps, and the question of who they belong to.
What people actually ask.
The two questions this page gets most, answered without a support ticket.
The words are ours. The brands are not, and nothing here is licensed.
The cast page names every person and what they are actually responsible for, which is not always what the programme implies.
