The one thing this service does
No satellite sees a spray. A spray is designed to leave the canopy unchanged, which is the point of applying it, and no instrument at any price detects droplets of mostly water on a canopy whose reflectance is dominated by chlorophyll and structure. The same holds for how much fertilizer went on. Good soil looks like good fertilizer from orbit.
What can be done is the inverse. A farmer reports spraying on a particular day, and the rainfall record over that plot shows heavy rain falling on it. Those 2 statements cannot both be true, and a contradiction like that is real, sourceable and worth acting on.
A flag rate, not a checkmark
This is the distinction the whole service rests on, so it is worth stating without hedging. We sell the flag rate. We do not sell confirmation.
Run a spray-date check across 1,000 farmers and almost every claim will come back consistent with the weather, because almost every day is sprayable. That agreement is not a result; the test passes for true and false claims alike. The small remainder that could not have happened is the finding. Read a flag as a farm worth visiting rather than as a conclusion about a farmer, and watch the rate across a cohort and across seasons, which is where it turns into a management signal.
Plot size, the cheapest check in the set
A farmer’s stated hectares against the area computed from their own polygon disagrees more often than anyone expects, and neither number is automatically the wrong one. Boundaries move, records are inherited, and a plot is sometimes described as the whole family holding when the field is only part of it. The disagreement is the useful output. It tells a programme which records to rebuild before the next season.
Pre-harvest interval
The check certifiers ask about most is arithmetic. Take the date a product was applied and the date the crop was picked, both given by the farmer and both timestamped, and the interval either meets the requirement or it does not. No imagery is involved, nothing is inferred, and the answer holds up under audit because it rests on 2 attested statements and on no measurement at all. Both dates come out of the farmer interviews, which is why cross-checks are always part of an interview study.