Every accounting system has one, usually populated automatically when you set up. The categories describe assets, liabilities, equity, income, and expenses, and every transaction lands in one of them. Your profit statement and your balance sheet are assembled by grouping those categories, which is why the chart determines what questions the reports can answer.
The default is designed to work for any business, which means it works particularly well for none. It typically bundles your largest costs into broad categories such as office expenses or general administration, and a business whose main expense is subcontractors or software will find the most important number hidden inside a bucket containing several unrelated things.
Split what you need to manage and leave the rest alone, which is the whole discipline. If subcontractors are a third of your costs, they deserve their own category and possibly several. If software is a growing expense you cannot account for, splitting it by purpose tells you where it went. Everything you rarely think about can stay in a general category, because a chart with ninety entries is a chart nobody files against consistently.
Separate direct costs from overhead, which is the single most useful structural decision. Direct costs are what it takes to deliver the work: materials, subcontractors, anything that rises with volume. Overhead is what the business costs to exist regardless. That separation is what produces a gross margin figure, and gross margin is the number that tells you whether the thing you sell is worth selling.
Set up income categories with the same care, and by service or product line rather than as one figure. A business with a single income category knows what it earned and not which work earned it, which means it cannot tell which service is actually profitable. That is usually the most valuable thing the accounts could have told you.
Keep it stable once set, because comparability across periods is the point. Renaming and reorganising categories mid year makes this year incomparable with last, and the historical comparison is where the useful patterns appear. Add a category when you genuinely need one, and resist restructuring.
Agree it with your accountant early rather than at year end. They will have opinions about what needs separating for tax purposes, and the categories that make your reports useful are frequently different from the ones that make filing straightforward. Settling both at the start avoids reworking a year of transactions.
Then check that transactions are actually landing in the right places, monthly, for fifteen minutes. Automatic categorisation is good rather than correct, particularly for suppliers you use for more than one purpose, and errors compound quietly across a year until the reports describe a business that does not exist.
Review it annually rather than never, because a business changes and categories that made sense at the start frequently describe a shape you have outgrown. Adding a category for something now material, and merging two that never held much, keeps the reports readable without disturbing the historical comparison.
Keep the number of categories small enough that filing is unambiguous, because a chart offering three plausible homes for the same transaction produces inconsistent records. If you hesitate about where something belongs, that hesitation will recur monthly and the resulting data will not compare against itself.