A chart of accounts that actually tells you something

Most chart of accounts we inherit were built by someone adding a line every time a new question came up. Twelve months later there are forty-one expense accounts, three of them are called "Software," and nobody can tell you which one the Stripe charges land in. The books balance. They just don't answer anything.

The problem is not the number of accounts

It is that the accounts were never designed around the decisions you actually make. A founder looks at a P&L and asks four things: what did we spend on people, what did we spend on acquiring customers, what did we spend on running the product, and what is left. If your chart of accounts cannot answer those four questions without a spreadsheet export and an afternoon, the chart is the problem, not the reporting.

The fix is not fewer accounts. It is a small number of parent categories with children underneath, so that the summary rolls up cleanly and the detail is still there when you need it. Think of it as a filing cabinet rather than a junk drawer. The top-level tabs should be stable for years. The folders inside can change as the business changes.

Start with five buckets, then subdivide

Every SaaS business we work with can be mapped onto five top-level expense buckets: people, customer acquisition, product and infrastructure, operations, and professional services. That is it. Everything else is a child. Payroll, contractor payments, and benefits live under people. Ad spend, content, and events live under customer acquisition. Hosting, third-party APIs, and tooling that touches the product live under product and infrastructure. Rent, insurance, and software the whole company uses live under operations. Legal, accounting, and advisory live under professional services.

The reason this matters is not aesthetic. It is that gross margin becomes a real number instead of a guess. If hosting and third-party APIs are scattered across "Software," "Office," and "Misc," your cost of goods sold is understated and your gross margin is flattering. When someone asks why the business looks profitable but the bank account does not move, the answer is almost always hiding in a misclassified account.

Name accounts for the question, not the vendor

"AWS" is a vendor. "Hosting — production" is an answer. Vendors change; the question does not. We rename accounts so that a reader who has never met the business can still understand what the money bought. That single habit removes most of the back-and-forth at year end, because your accountant is not calling to ask what "MISC 2" was in March.

The same rule applies to revenue. One revenue account called "Sales" tells you nothing about whether you are growing subscriptions, one-time implementation work, or overage. Splitting recurring subscription revenue from services revenue is the difference between knowing your real growth rate and guessing at it. It also makes it obvious when a good month was actually a services month wearing a subscription costume.

Give every account a job and retire the rest

An account that has had no activity in twelve months is not a safety net, it is clutter that makes the active accounts harder to find. We archive them. Not delete, because history matters, but remove them from the picker so nobody accidentally codes a transaction to "Travel — 2019 conference" again. The test is simple: if a bookkeeper cannot pick the right account in under three seconds, the list is too long.

Do this once, then leave it alone

The temptation is to redesign the chart every quarter. Resist it. A chart of accounts is only useful if it is comparable across months. Changing structure mid-year means every trend line breaks and every year-over-year comparison becomes a research project. Set the structure, document what belongs where in one page, and change it at a fiscal year boundary if you must change it at all.

If you are not sure whether your current chart passes the test, pull last month's P&L and try to answer the four questions in under five minutes. If you cannot, the structure is costing you more than it looks. That is usually a two-hour fix with a lasting payoff, and it is the kind of thing worth doing before the next twelve months of reporting pile up on top of it.