Check missing and conflicting data

A stock equation may balance while some inputs remain unsuitable for a decision. A supplier invoice may cover two crops without allocation, a quoted price may refer only to premium grade while the quantity includes lower grades, or a rainy day's harvest may be estimated from unweighed crates. Before budgeting, classify the **evidence status** of each material number: directly measured with a record, calculated from measurements, explicitly estimated, or disputed. These labels describe data quality rather than anyone's honesty. A carefully bounded estimate can be more useful than an impressive decimal figure whose meaning is unknown.

Keep an assumption register alongside the worksheet. For each important value, record its unit, period, source, observation date, calculation method, person who can clarify it, and plausible range if uncertain. For example: “Tomato packaging cost, 900 units on a shared tomato-cucumber invoice; estimated tomato share 60% using crates issued.” Neither charge all 900 units to tomatoes nor present 540 as a measured fact. Look for an actual crate-issue record. If a provisional choice cannot wait, run both 50% and 70% allocation assumptions and ask whether the ranking changes.

Distinguish **missing** from zero. A blank water-cost cell may mean the pump invoice has not arrived, not that irrigation is free. An intentional zero means the item is outside the decision boundary or borne elsewhere, and that treatment needs explanation. Do not take an old market quote to fill a current-season sales-price gap and label the result an actual sale. If a planning estimate is needed, state its date, grade and delivery point. Before resolving apparently conflicting records, align their definitions: gross box weight and net product weight can both be correct. Keep disputed values and their sources visible until reconciliation rather than silently replacing one.

Prioritize missing data by their effect on the decision, not by how easy they are to obtain. If the crop choice is unchanged throughout a reasonable packaging-cost range, record the assumption and proceed with a bounded recommendation. If a small change reverses the choice, obtain the measurement or invoice before a commitment. A simple range test can therefore be more valuable than filling every blank with an arbitrary point estimate. Escalation is proportionate to the exposure and the authority required to accept the uncertainty.