Rules
Part of Chart of accounts 2027? Five types and what gets a line
Chart of accounts metrics: the numbers that show if it still works
Chart of accounts metrics you can compute from a ledger export: dormancy, thin accounts, container share, coding consistency and the age of unresolved items.
You can measure an account list. That surprises people, because a chart feels like a matter of taste, and every argument about it tends to end in someone asserting that their version is tidier.
It is not taste: a chart's observable properties include how much is used, how much value it concentrates, whether the same transaction reaches the same place twice running, and how long unclear items sit before somebody resolves them.
Count all of it from a ledger export in about an hour. The counts tell you what to merge, split, or close.
Every figure below is invented to show how the calculation runs.
What to take away
- Count usage, not accounts. A list of 200 where 180 are used is fine; a list of 60 where 20 are used is not.
- The two numbers that matter most are the share of value sitting in vague containers and the age of the oldest unresolved item.
- Measure once a year from a ledger export, write down what you found, and compare next year. A single reading tells you much less than a trend.
The measurements worth taking
| Measurement | How to compute it | What a poor reading means |
|---|---|---|
| Dormancy | Accounts with no entries in twelve months, as a share of all accounts | The list describes a business you no longer run |
| Thin accounts | Accounts with fewer than about a dozen entries in the year | Coding decisions being made repeatedly for a line nobody reads |
| Container share | Value in accounts named general, other, sundry, miscellaneous or similar, divided by total expense value | Costs you have decided not to understand |
| Coding consistency | For your twenty largest suppliers, how many distinct accounts each one landed in | The definitions are missing or contradictory |
| Reclassification volume | Value of journals whose only effect is moving between accounts on the same statement | Coding is being done twice, once wrong and once by whoever cleans up |
| Holding age | Age of the oldest item in the suspense or unclear account, and value cleared within thirty days | The question loop is not closing |
| Read-through | On last month's report, lines you skipped, plus lines you had to open the ledger to understand | Too many accounts in the first case, too few in the second |
| Structure churn | Accounts added, merged or retired in the year, and how many have a documented mapping | Comparatives are breaking without anyone recording why |
Working two of them through
Thin accounts, and what they cost. Suppose an expense list of 140 accounts. Pulling the year's entries shows 46 accounts with fewer than a dozen postings, and together those 46 hold 3% of total expense value.
Merge thin accounts, keep the required ones
Merge
- Reason
- Nobody reads the line
- Test
- Under a dozen entries
- Effect
- Shorter reports, faster coding
- Action
- Merge at period boundary
Keep
- Reason
- External party asks
- Test
- Covenant or funder condition
- Effect
- Split stays justified
- Action
- Note why it survived
Thin accounts: 46 of 140
- 140expense accounts in the list
- 46accounts with under a dozen postings
- 3%of total expense value they hold
- 1coding decision needed each time
That 3% is the argument, and those 46 accounts need a coding decision each time a transaction that might belong in one appears. They appear on every report as lines nobody reads, and the total they separate out would not change a single decision.
Merging them into reporting groups shortens reports, speeds coding, and leaves the information unchanged.
Watch for the exception before you merge. An account can be thin and still necessary when someone outside asks for it. A lender's covenant, a funder's condition, or a required report justifies a split your own management would not need. Anyone likely to approach the ordinary routes to business funding should find out those conditions before merging anything.
Merge everything thin except those. Note beside each survivor why it survived.
Container share, and the arithmetic that makes the case. Say total expenses for a year are 480, and the accounts called general expenses, other costs and sundries hold 34 between them. That is 7%.
Seven percent does not sound alarming until you put it beside a real line. If operating profit that year is 38, the uncategorized value is nearly the size of the result you are explaining.
Any question about why the profit and loss statement moved has a 34-shaped hole in the middle.
The fix is not a rule against vague accounts, which nobody follows. It is a monthly habit: read the container accounts every month, and when the same kind of thing appears in one three times, give it a real account with a written definition.
The consistency check that finds the most
Take your twenty largest suppliers by value. For each, list the distinct accounts their transactions landed in over the year.
The supplier consistency check
- Take your twenty largest suppliers by value
- List distinct accounts each landed in
- Flag one-product suppliers in four accounts
- Write a definition naming which account wins
- Prevent the next twelve months, not the past
Some spread is legitimate. A general merchant genuinely sells you three different kinds of thing. What you are looking for is the supplier who sells one thing and appears in four accounts, because that is a definition problem, and it means every report containing those four lines is describing coding habits rather than spending.
The check's output is not a correction. It adds sentences to your account definitions, each naming which account wins an ambiguous case.
Correcting history is optional; preventing the next twelve months is not. Those definitions are the artifact a chart derived from outputs depends on, as set out in the framework for building a chart backwards from reports.
Measurements that look useful and are not
- The number of accounts on its own.It is meaningless without usage and without the size of the business. A count is only interesting as a trend against your own past.
- Comparison with another business's list.Their splits answer their questions. The similarity tells you nothing about whether yours answer yours.
- How closely the chart matches a system template.A template is a starting draft. Matching it perfectly after two years of trading means nobody has adapted it.
- Depth of the hierarchy.Deep nesting looks organized and often just moves the problem: five levels of subaccount is still a coding decision every time, made harder by the scrolling.
What to do with the readings
Turn each into an action with a date rather than a score.
Five actions with dates
- Merge thin accounts at a period boundary
- Promote container items to named accounts
- Write a definition line per flagged account
- Retire dormant accounts by stopping entries
- Cap holding account at one month
Then take the same readings a year later. Container share and holding age move fastest and are the best early signal that the discipline has slipped. The structural rules underneath these actions are in the chart of accounts overview, and the errors these readings usually surface are cataloged in the mistakes that damage a small set of books.
