Historical farm data
We track both alive and dead animals across production, milk-test and herd-event records over time to analyse the rate of change.
See the risk behind the headcount
MooNitoring brings your livestock book into one map: where risk is concentrated, and how it changes farm by farm, so you can select, price and monitor each farm on its own history.
MooNitoring Farm Risk Map
The portfolio average hides the differences
The map follows every farm in the portfolio over time. With one shared rate, they all receive the same risk expectation.
One rate for every farm
45.2500° N · 10.0500° E20 km
2025 REPLAY · JAN–AUG
Why this works
Declarations describe only a point in time
Farm systems instead record production, milk tests, lifecycle events, treatments and exits over months and years. That sequence can show farm differences that a shared average misses.
It can also reveal what happened before a severe event, without claiming that the data proves its cause.
Real farm data
MilkingDaily
ActivitySignals
HealthEvents
Milk testsLab
One cow identity
Dates, source references and derivation rules stay attached.
Clear insights
UnderwritingSelection
MonitoringEarly warning
Event reviewTraceable
The evidence
Farm-specific models beat one shared average
We compared predictions for six farms in 2025 with predictions based on the portfolio average. Shorter bars mean more accurate predictions.
Mortality
Average miss, head of cattle
−46.5%46.5% less error
on 5 of 6 farmsChronic mastitis
Average miss, head of cattle
−68.6%68.6% less error
on 6 of 6 farmsProduction
Average miss, kg per head per day
−66.7%66.7% less error
on 6 of 6 farmsLower prediction error on each farm means better selection, better pricing, and earlier warnings. That is how we expect to help you improve loss ratios.
This is possible thanks to 73 cow-level features that we map and combine using our proprietary models.
−60.6%60.6%prediction error on average, across the six farms and three targets above.
| Target | Unit | MooNitoring | Baseline | Reduction | Farms with lower error |
|---|---|---|---|---|---|
| Mortality | Average miss, head of cattle | 21.58 | 40.31 | 46.5% | 5 of 6 |
| Chronic mastitis | Average miss, head of cattle | 13.95 | 44.45 | 68.6% | 6 of 6 |
| Production | Average miss, kg per head per day | 1.46 | 4.40 | 66.7% | 6 of 6 |
The data product
What you can test against your book
Consistent definitions
Our technology processes and transforms data from different software and farms into a single source of truth.
Traceable values
Records retain their source, date, animal identity, derivation rule and quality status so that you can verify the original data.
Fraud detection
We combine multiple domains of the dairy science and machine learning to detect fraud.
How we can work together
Three phases, one clear answer
Working session
We define the goal together: the risk or loss you want to see differently, and the decision it would change. You bring the underwriting perspective; we bring our domain knowledge.
Retrospective test
We contact the farmers and run the research against their dataset. Since no new hardware is involved, we can manage everything remotely.
Review
We show you what the data supports and how it can improve the underwriting. If it does not support your use case, we say so in plain terms.
Bring us a hard livestock question!
Book a sessionFAQ
Questions insurers actually ask
What does MooNitoring deliver to an insurer?
Dairy models to test against your portfolio. The approach is designed to help insurers improve loss ratios by assessing each farm on its own history.
What happens in a first working session?
We understand what your needs are and how we can help you, and set the success criteria.
How long does a retrospective test take?
We scope the timeline in the working session and commit to it there. It depends on the question, so we do not publish a number we might miss.
How were the model results validated?
A repeated temporal backtest for 2025: models selected and trained only on data through 2024, forecasts sealed, then compared against observed outcomes on six farms.
Which markets are covered?
Connected farms are Italian but the data model is designed to extend across European and US dairy markets.