Before you read a number from the farm’s work for the first time, write down three things: the exact source it comes from, how often it is read, and the name of the person who reports it. That is what lets someone else check it. Choosing the number is the easy part; this writing is the real work, and it has to happen before the first reading, not after. The three things fit on one row of the notebook or spreadsheet where you keep the number, and no hour meter or farm software fills them in for you.
On most farms the raw numbers already exist, but they are scattered. Machine hours sit in a notebook in the shop, a photo of the hour meter waits on a phone, a figure the foreman carries in his head comes out when someone asks. Then a decision arrives (replace the machine, change the work schedule, question how a field job was done) and nobody can say where the number came from, how current it is, or why two people quote it differently. What these numbers lack works like the address on a letter: where each one comes from, how often it is read and who reports it.
Why do two readings of the same number disagree when nobody wrote down the source?
A number is not a fact waiting in the field, it is the output of a measurement, and a measurement nobody defined can be taken two ways. To check a number, a farm-data guide from the FAO, the United Nations food and agriculture agency, asks where it came from: is it clear where the figure came from, and whether it changed before you use it. The same guide warns that the unit has to be fixed at the point of collection or the figures stop being consistent with each other. Two readings of the same operation, taken by two people against two unwritten methods, are two different numbers that happen to share one name.
The trap is a number that looks fixed and is not. Take days open in a dairy herd, the days from calving until the cow is pregnant again. In a survey of dairy consultants in the journal Frontiers in Veterinary Science, Armengol and colleagues note that this figure can only be worked out once a pregnancy is confirmed. Yet some herd software treats a cow not yet confirmed pregnant as if she had conceived on the day the report is printed, which gives a false reading of the herd. Same indicator, same name, two ways of counting, two answers. When the source is written down, you can see which count you are looking at before you compare them.
Which one number should you pick, when everything can be counted?
Pick the single indicator that already feeds a decision you make, not the longest list you can keep. A Scientific Reports study of UK sheep systems found that livestock advice has piled up an overabundance of ambiguously defined indicators, that this glut discouraged farmers from recording beyond a minimum, and that the duplication invited uncertainty about what each one is even for. Measuring is not free either: the same authors note that gathering the information is seldom without cost. One number read and checked beats ten written and trusted.
Picking it is a judgment call, and no standard list makes it for you. The dairy-consultant survey found that the farm’s size and the advisor’s experience changed which indicators they reached for, and that a short set was enough for a first pass. The number worth writing down this way is the one whose movement would change what you do next. Recording the others is collecting data for its own sake.
Where does the number actually come from?
Every number comes from a specific reading, and declaring the source means naming that reading exactly. Take a machine’s work rate, the area it covers per engine hour, in acres or hectares. It comes from two readings, the hour meter and the mapped area of the field, and the rate changes with what you count as working time. A machinery-cost guide from South Africa’s agriculture department builds that into what it calls a field-efficiency factor, which subtracts the time spent turning at the headlands and refueling the tractor, and in one worked case that lost time came to 17 percent of the working time. Count that time and you get one rate, leave it out and you get another.
So the source line is not “hour meter.” It is which meter, on which machine, over which field, with turning and refueling time counted or not. A number whose method is unwritten cannot be set beside next month’s, because next month someone may count it the other way, and the two will disagree for a reason nobody recorded.
How often is it read, and who reports it?
A single reading is only true for the moment it was taken, so how often it is read is part of the number, not something added later. The University of Maryland farm-records guide makes the point about any count: it is valid only for the point in time it was taken, because the thing keeps changing, and to track a resource over time the same guide says it has to be inventoried regularly. A number read once a year and the same number read every week are two different measurements, and the row has to say which one you have.
Someone has to be named, or the reading does not happen. The frequency answers when, and the reporter answers who takes the reading and whose reading stands when two disagree. That name is the difference between a number written where anyone can read it and a number that lives in one person’s memory until that person is on the other tractor.
What do you write in the row for the number?
One row, five cells: the number itself, the three things named above, and the first reading with its date. A person has to decide or write each cell; the machine alone fills none of them:
| Cell | What it names | Filled example |
|---|---|---|
| Indicator | the one number tracked | acres per engine hour, planter |
| Source of the number | the exact reading, and what counts in it | hour meter on tractor 2, area from the field map, turning time included |
| Frequency | how often it is read | after each planting pass, written up the same evening |
| Who reports | the person who takes the reading | the operator, checked by the foreman |
| First reading | the number, with a date | the number read today, written with today’s date |
Read down the second column and none of it comes from pressing a button. The machine gives a raw count. Writing the row is what turns that count into a number another person can pick up and trust.
When can a number with an address still mislead?
A number with its source, frequency and reporter written down can be checked, which is not the same as being correct. Writing them down gives you a number you can argue with; it does not make it the right number to watch. The Scientific Reports work is blunt that many recommended indicators carry little information when used on their own, and watching a number also tempts whoever reports it to manage it to look good, which no row can prevent. The method has to be revisited, not set once and forgotten.
Each part of the row is a choice someone made: what the reading includes (turning time in or out), how often is often enough, who is trusted to read it. Writing those choices down lets the next person see them, instead of inheriting a number with no reason behind it. Writing the row does not remove the judgment; it puts the judgment where others can see it.
Where to start
It takes an afternoon, with what you already measure. If you plan field work on a written calendar, with the earliest start and latest finish for each operation, write the row on the same page, next to the job the number measures. The same habit of naming where a figure comes from works elsewhere too: it lets you check an input count against its record in the shed, and read the hour meter against the service interval printed on the machine. More numbers worth planning around are in the planning library, and the guide to managing the farm as a business shows how a number like this feeds the decisions you make across the farm.
A number with an address can be handed to someone else and survive the handoff. A number without one can only be taken on trust, and farms already run on plenty of numbers like that: the horsepower everyone quotes, the yield everyone remembers, the cost everyone assumes. The first number you give an address to is the first one that stops depending on you being in the room.