Ask five different people how long a farm record has to be kept and you will get five different numbers, and every one of them will be right for the record they were thinking of. The tax authority wants one window. The organic certifier wants another. The buyer who signed a supply contract wants a third. None of them is wrong, and none of them is the answer for every record on the farm, because there is no single answer. There is only a method for finding one, record by record.
This page is that method. It will not tell you a number of years to write on a shelf label, because any number it gave you would be wrong for some record, in some jurisdiction, under some contract you have not read yet. What it will give you is the process for working out, for your own records, which obligation reaches back furthest — because that is the one that decides.
The authority is never this page. Retention requirements come from your tax authority, your certifier, your regulator, your insurer, and the contracts you have signed with buyers and lenders. They differ by jurisdiction, by certification scheme, and by the specific record in question, and they change. What follows is the process for identifying and combining those requirements — not a substitute for asking each authority directly what its window is.
A retention schedule is a table of authorities, not a table of years
The useful version of a retention schedule does not start with a time period. It starts with a list: every party who might, at some point, have a legitimate reason to ask you for a given record. For a spray record, that list includes your organic or food-safety certifier, the regulator who licenses the product you applied, and any buyer whose contract requires proof of application history. For a payroll record, it includes a labor regulator, a tax authority, and the employee themselves. The record does not have one owner. It has a list of people who can ask, and each one brings their own window with them.
Once the list exists, the period follows almost mechanically: find out, from each party on the list, how far back their own rules reach. You are not being asked to guess. You are being asked to go and find out, once, and write the answer down next to the record type — not as a policy you invented, but as a citation to the party who told you.
The longest applicable period governs, always
This is the single load-bearing rule in the whole subject, and it resolves almost every question that feels complicated in the abstract. If three different authorities have three different windows for the same record, you do not average them, and you do not pick the one that is most convenient. You keep the record for as long as the longest of the three requires, because satisfying the longest window automatically satisfies the shorter ones nested inside it.
The practical effect is that a farm rarely needs a different retention period for every record type. It needs a small number of tiers — short, medium, and long — and a habit of asking, for any new kind of record, which tier its longest obligation belongs to. Records tied to a residue or a certification claim tend to sit in the long tier, because an investigation or an audit can reach back further than a routine business record usually needs to.
This is worth naming for the record it affects most: a withdrawal record often answers to the longest of several look-back windows at once — a certifier's, a buyer's, and a residue investigator's — which is exactly why its own retention is usually the long tier on any farm's schedule, regardless of the species or the product involved.
The clock does not start on the same day for every record
A subtler failure than picking the wrong length is picking the wrong start date. Some records are governed from the date they were created — a purchase invoice, say. Others are governed from the date an event they describe was resolved — a treatment record whose clock, for some purposes, does not really start until the affected animal or lot has left the farm. Others still are governed from the end of a season or a contract, not from the date of the entry itself.
Get the start date wrong and a record can look retired years before its actual obligation expires. This is worth writing down explicitly next to each retention tier — not just "how long" but "counted from what" — because the second question is the one that quietly gets skipped.
A schedule has to work without you in the room
A retention schedule that lives only in your memory is not a schedule — it is a habit that ends the day you are unavailable to enforce it. The useful version is written down somewhere anyone on the farm can find it: which record types exist, which tier each one belongs to, and what triggers the clock. This matters more than it sounds, because the person who eventually deletes an old folder to make room, or throws out a box because the shed is being cleared, is very often not the person who understood why it needed to stay.
The same discipline that makes a farm recordkeeping system work day to day — a place everyone writes, a form everyone uses — is the discipline that keeps a retention schedule alive past the season it was written in. It also depends on the records themselves being labeled consistently enough to sort by tier in the first place, which is really a question of naming as much as of retention.
Destruction is a decision, not a default
The end of a retention period is not an instruction to delete anything. It is the point at which you are free to, if space or storage cost requires it — and even then, only after checking that no newer obligation has extended the window since the schedule was last written. A certifier can change its look-back requirement. A new contract with a buyer can impose a longer one retroactively for records you are still creating. Treat the expiry of a retention period as a prompt to re-check the schedule, not as a green light to clear the shelf.
This is one of the twelve one-click CSV exports Farm40 provides for every record type: a durable, dated copy you can move into long-term storage once a record's active use on the farm is done, without needing the live system to still be running years later. The limit is worth being direct about — an export is a snapshot of what you entered at the time you took it. It cannot tell you that a record was never created in the first place, and it does not know, on its own, which retention tier a given export belongs to. That judgment is still the schedule's job, not the export's.
A retention schedule, done honestly, is unglamorous work: a short list of record types, an authority next to each one, a start date rule, and a tier. It is also one of the few pieces of farm recordkeeping that costs almost nothing to get right and can be genuinely expensive to get wrong, precisely because the record you need is always the one you already deleted.