You can make a good call and still get a bad year.
That is not an excuse — it is why the quality of a decision has to be judged separately from the outcome. Farming is a noisy decision environment: weather, biology, currency, international trade — you cannot control all of that. You can control the process you use to make the call.
Good decision-makers are not always right. They use processes that improve expected outcomes over repeated decisions.
We went looking outside agriculture for this — aviation and human factors, high-reliability organisations, forecasting research, behavioural economics, naturalistic decision-making, evidence-based management. Across all of them, the same pattern keeps appearing. Better decisions come from a repeatable process: clear objectives, a current view, sound mental models, challenge, and a loop that learns from what happened.
Judged by the process and the information available at the time
Was the objective clear? Was the picture current? Were the plausible futures weighed rather than assumed? Could the decision have been made better with what was knowable that day?
Judged by what the season actually did
Noisy, lagging, and partly luck. A sound call gets punished by a drought; a poor one gets rewarded by a schedule lift. Judge yourself on this alone and you learn the wrong lessons in both directions.
Ten things that separate repeatable decision quality from luck.
Drawn from decision science, human factors and organisational learning research. They reinforce each other — a clear objective improves evidence selection; evidence discipline improves causal models; probabilistic thinking improves premortems; debriefs improve future intuition.
Clear objectives
What are we actually optimising for?
Situation awareness
What is really happening now?
Causal mental models
Why is it happening, and what changes what?
Evidence discipline
What evidence should inform the call?
Probabilistic thinking
How likely are the plausible futures?
Cognitive reflection
What might my first answer be missing?
Earned intuition
Is this a domain where gut feel is trustworthy?
Constructive dissent
Who can safely challenge the view?
Premortem and option design
How could this fail, and how do we keep options open?
Learning loops
How do we turn outcomes into better future judgement?
This is why the product behaves the way it does.
Each involved a trade-off in the build. We chose them because the research points towards the same thing again and again: better decisions come from visible assumptions, explicit uncertainty, cheap challenge, and feedback from what happened — even when the alternative looks better on a screen.
Ranges, not single numbers
Driver 05 · probabilistic thinking
A single confident figure is easier to design and easier to sell. It also manufactures certainty that isn't there. farmBI answers with a band, and the band gets wider when your data is thinner rather than the tool refusing to answer.
Provenance on every figure
Driver 04 · evidence discipline
Measured, entered, or a starting default — the screen says which, and you can change any of them. A number whose source you can't see can't be weighed against what you already know.
Shows the trade-off, not the decision
Drivers 01 & 07 · objectives and intuition
farmBI does not tell you the best option, because it cannot know what you're optimising for and it has not stood in your paddock. It shows what changed and what the trade-off is. The call stays yours.
Cheap to re-run
Drivers 06 & 09 · reflection and option design
If asking a second question is expensive, nobody asks it. Scenarios re-run in seconds so the pessimistic version gets tested as a matter of habit, not as a special exercise.
The month gets closed
Driver 10 · learning loops
Counting up and comparing it to what you expected is the least glamorous feature we've built and probably the most valuable. Without it, too much of what the season taught you gets lost before the next one.
Ten cross-industry drivers of repeatable decision quality.
The full brief reviews decision analysis, human factors, high-reliability organisations, forecasting, behavioural economics, naturalistic decision-making and organisational learning — with the observable behaviours and failure mode for each driver.