The twenty-odd categories every transaction lands in. Get this right once and every report you ever run is useful; get it wrong and no amount of software helps.
12 min read · Level 2
Your chart of accounts is the set of questions you will be able to answer for the rest of the company's life. Not the questions you can ask — the ones you can answer, instantly, from records that already exist. Founders treat it as admin and let software pick the defaults, then spend two years unable to say what their gross margin is because hosting and salaries went into the same bucket. Twenty minutes of thought here buys clarity you cannot retrofit.
The chart of accounts is the list of categories transactions get recorded into, grouped the way the income statement reads:
Revenue → COGS → Operating expenses → and, on the balance sheet, assets and liabilities.
For a small company, fifteen to twenty-five lines total.
Because every report is a re-slice of these categories. If you never separated hosting from salaries, you cannot compute gross margin — not with better software, not with an accountant, not ever, without re-categorising every historical transaction by hand.
And because the split you most need is the one software defaults get wrong: which costs are COGS and which are operating expenses. That single boundary is what produces gross margin, which is the number that says whether your business model works.
A workable chart for a small software company:
Revenue
Cost of goods sold (rises with each customer)
Operating expenses (does not rise per sale)
About eighteen lines. Enough to answer every question a founder actually asks; few enough that categorising stays consistent.
The two decisions that matter most are both in there: support salaries sit in COGS because they scale with customers, and engineering salaries sit in OpEx because they do not. Move those and your gross margin changes by tens of points.
First-time founders accept the software's default chart, which is usually generic and has one giant "General expenses" bucket. Six months later half the spending is in it and no report is informative.
The second mistake: too many categories. Forty lines means nobody categorises consistently, including you, and inconsistent categories are worse than coarse ones because they look precise.
The third: putting all salaries in one line. Engineering, support and sales salaries behave completely differently — one is OpEx, one is COGS, one drives CAC — and merging them destroys three metrics at once.
Revenue, COGS, OpEx. Every category belongs to exactly one of the three, and knowing which is the whole exercise.
Yes → COGS. No → OpEx. That test settles almost every case.
Engineering, support, go-to-market at minimum. They answer different questions and merging them is irreversible without re-categorising history.
Enough resolution to be useful, few enough to stay consistent.
One sentence: "contractor work delivered to a customer is COGS; contractor work on the product is OpEx." Future-you will not remember.
Changing categories mid-year breaks comparison. Batch changes at the year boundary.
Set it up in your first bookkeeping session, before importing transactions.
Do not redesign the chart every quarter chasing a cleaner report. Comparability across months is worth more than perfect taxonomy.
Where the COGS line sits is one of the most consequential accounting judgements a subscription company makes, and it is genuinely contested — companies differ on whether customer success, professional services and infrastructure sit above or below it. Investors comparing two companies with the same product can see very different gross margins purely because the boundary was drawn differently.
Pick your boundary deliberately, write down the rule, and apply it consistently. A defensible 72% beats an undefendable 85%.
A common sequence: default chart accepted, everything lands in two or three broad buckets, and the founder cannot answer "what is your gross margin?" in a fundraising conversation. Re-categorising eighteen months of transactions to answer it is typically several days of work, done under deadline, in the middle of a raise.
The question you cannot answer at seed is nearly always gross margin. Build the chart that answers it before anyone asks.
Create a free account to read the rest and apply it to your own startup.