Direct answer first: keep the records that prove your business to the people legally entitled to ask — transactions, assets, liabilities, ownership, contracts, and payroll — organised so that any of them can be shown in one step. The standard is simple: if a tax authority, auditor, lender, or investor asked for the paper behind a number today, could you produce it this afternoon? This guide covers the five record families, the showability standard, structure versus volume, digital records, and retention — with the honest note on retention periods.
The record families
Records are not a pile; they are five families, and each family answers a different question about the business:
1. Transaction records
Invoices issued and received, receipts, payment records, bank and card statements. These prove what happened: what was sold, what was bought, what moved. They are the family that tax filings are built from, which makes them the family that queries start with.
2. Asset and liability records
Fixed assets and their depreciation, loans and their schedules, leases, guarantees, and the documents behind them. These prove what the business owns and owes — the family that a balance sheet, a lender, or a buyer reads first.
3. Ownership and governance records
Shareholder registers, board minutes and resolutions, the incorporation documents, and the decisions that shaped the business. These prove who owns it and how it is run — the family that changes hands, lawyers, and due-diligence teams ask about.
4. Contract records
Customer agreements, supplier contracts, employment agreements, NDAs, and the correspondence that goes with them. These prove what the business promised — the family that disputes and renewals read.
5. Payroll and people records
Employment contracts, payroll runs, tax withholdings and contributions, leave and termination records. These prove what the business did for the people it employed — the family with its own regulators and its own questions.
The five families are worth memorising for one reason: when someone asks "do we keep this?", the answer is found by asking which family it belongs to — and the answer is almost always yes, organised under that family.
The standard: could you show it?
The retention question is simpler than it sounds if you replace "how long do we keep it?" with the standard that actually governs everything: could you show it? Every record family exists to be shown to someone with a right to ask — the tax authority asking about a filing, the auditor asking about a balance, the lender asking about a customer contract, the regulator asking about a hire. The practical standard is therefore not a shelf-life in years; it is an answerability in minutes: any record that the business's obligations might require it to produce is kept, organised, and retrievable for as long as that obligation can bite.
Two consequences follow. First, the records with the longest "show" horizons are the ones tied to long-lived obligations — ownership, governance, contracts, and the tax records behind filings whose review windows extend well beyond the filing itself. Second, the records with no show-horizon at all — the email that led to nothing, the draft that was superseded, the receipt for an expense that was never claimed — are exactly the ones you can discard with a clear conscience. Showability gives you both halves of the retention decision: what to keep and what is safe to let go.
The five families at a glance
| Family | What it proves | Who asks to see it | Retention logic |
|---|---|---|---|
| Transactions | What happened — sales, purchases, payments | Tax authority, auditor | As long as filings can be reviewed |
| Assets and liabilities | What the business owns and owes | Lender, buyer, auditor | For the life of the obligation |
| Ownership and governance | Who owns it, how it is run | Buyer, lawyers, registry | Permanently |
| Contracts | What the business promised | Dispute parties, renewals | While the promise can bite |
| Payroll and people | What it did for employees | Labour and tax regulators | Per employment/tax rules |
Read the "retention logic" column and the pattern is clear: the family's shelf-life is set by the longest-lived obligation that can ask for it. That is the principle behind the retention section below — and it is why "keep everything forever" is as wrong as "keep it for three years": neither matches the shape of the actual obligations.
Structure beats volume
The businesses that pass record requests in an afternoon do not keep more paper than the ones that fail — they keep the same paper in a structure. Three structural decisions do all the work:
- One home per family. Each of the five families lives in one place — a folder structure, a drive, a cabinet — with the same naming in every place. The single most expensive habit in small-business record-keeping is the three-drawer scatter: some paper in the office, some in email, some on the bookkeeper's desk.
- The filing rule: file at the event. The document goes into its family at the moment it exists — the invoice when it is issued, the contract when it is signed, the statement when it arrives. Filing "at the event" is the difference between a records system and a records project. A month of receipts filed at the event is ten minutes; a year of receipts filed at tax season is an archaeology dig.
- The one-step test. Any number in any report or filing must be traceable to its source document in one step. If you have to remember where a document lives, it is not organised; it is merely somewhere.
Pick any figure in last month's profit report — a sale, an expense, a tax charge — and ask: where is the paper? If you can walk to it in one step, the structure works. If the answer is a hunt, the structure is the problem, and no amount of storage will fix it.
Records at three scales
The same five families look different as the business grows, and it helps to know what each scale requires:
- Solo or micro: one drive, one folder per family, filing at the event, a monthly sweep of the inbox into the folders. At this scale the system is the habit — ten minutes a week keeps the structure true.
- Small team: the families gain an owner each (the office manager owns payroll records, the accountant owns transactions), and the structure moves to the team's shared space with clear naming. The risk at this scale is the split-brain: documents filed in personal folders instead of the shared home.
- Multi-entity or multi-jurisdiction: each entity gets its own complete set of the five families — never mixed — and the structure becomes a formal rule (naming, versioning, access rights) rather than a habit. The audit trail question becomes real at this scale: records that can be changed without trace are a liability, and the structure should prevent it.
The scale matters because the failure mode changes: at micro scale the failure is a habit lapse; at team scale it is scatter; at multi-entity scale it is mixing and editability. The five families and the one-step test stay identical — only the enforcement changes.
Digital records: good news, with rules
Digital records are fully acceptable everywhere that matters — the requirement is not paper, it is reliability. The rules that make digital records work are simple, and breaking them is how businesses lose history without noticing:
- Complete and legible. A scan of the first page of a two-page contract is not the contract. The digital copy must be the whole document, readable now and in ten years' format migration.
- Unchanged, or change-tracked. A record that can be quietly edited is not a record — it is a claim. The practical minimum is a structure that prevents or tracks changes: locked PDFs, versioned folders, or a system with an audit trail. The businesses that learn this the hard way are the ones whose "records" were spreadsheets that anyone could amend.
- Retrievable and backed up. The record must be findable in one step (the structure above) and must survive a laptop, a drive, or an email account dying. The backup rule is simple: if losing one device would lose a record family, the family is not backed up.
- One system, not inboxes. The risk of digital record-keeping is not that records disappear — it is that they scatter across inboxes, drives, phones, and apps, so that "where is the paper?" becomes a five-app treasure hunt. The one-home-per-family rule applies to digital records twice as hard.
Retention: the honest answer
The question everyone wants a number for — "how long do I keep records?" — has an honest answer and a practical one. The honest answer: retention requirements differ by jurisdiction and by record type, and anyone who quotes you a single number for all of them is simplifying. Tax authorities typically set minimum retention for tax records that outlast the review window for filings; company law typically sets its own horizon for accounting records; employment and contract law set others for their families. These are real differences, and they matter.
The practical answer, in one paragraph: keep the ownership and governance family permanently; keep the transaction, asset, liability, and payroll families for at least as long as the jurisdiction's tax and company retention rules require — and when in doubt about a specific document, keep it until the obligation it supports can no longer bite. Then confirm the actual numbers for your jurisdiction with your advisor rather than a remembered figure — this is a ten-minute conversation that saves a decade of doubt.
The one habit that makes retention painless regardless of the rules: the structure above. A family that is filed at the event and housed in one place costs nothing to keep an extra year; the cost of records is never storage — it is disorder.
Records as the compliance backbone
Every compliance obligation in the calendar depends on the records layer underneath it: the filing is assembled from the transaction family, the renewal is evidenced by the governance family, the query is answered from whichever family it touches. A compliance calendar with no records structure is a schedule with nothing scheduled to back it — the obligations will be met in the sense of "done" and fail in the sense of "shown".
This is why the records review belongs in the quarterly rhythm: one item per quarter — pick a family, test the one-step rule, confirm the structure held. A quarterly ten-minute records check is the cheapest insurance in the whole compliance system. And when the records are questioned — by a tax authority, an auditor, a lender, or a buyer — the difference between an afternoon and a crisis is entirely the structure you built before the question existed. Our compliance calendar shows where the records layer sits in the obligations rhythm, and the tax-readiness calendar covers why the records must be filing-ready rather than reconstructed.
The bottom line
Business records are five families, one structure, and one standard: could you show it? File at the event, house each family in one place, keep the tax, ownership, and governance families for the full retention horizon your jurisdiction requires — and test the one-step rule quarterly. Do that, and the records stop being a pile and become the quiet backbone that makes every filing, query, and review routine.
