Analytical dimensions, not another account

You want to know how each site is performing. Or each project. Or each channel.

The reflex I run into everywhere is the same: create accounts for it. Revenue site A, revenue site B, purchases project X. It works, in the sense that the numbers land somewhere.

And then within two years you have a chart of accounts nobody can read any more.

Why that jams up

Splitting by account multiplies. Three sites times four channels is twelve variants of every account you want to split. Add a site and you have to create a new account for every existing one, and from that moment your history is no longer comparable without manual work.

The second problem is worse: you can only split one way at a time. If your accounts are per site, you cannot simply look per channel. And if you wanted per site and per channel, you are back at those twelve.

And the third: your accountant has to work with it. Every entry demands a choice from a list that keeps getting longer, and one wrong choice lands straight in your result.

What analytical dimensions do differently

An analytical dimension is a label you hang next to the entry, separate from the general ledger account.

You simply post to revenue. And you attach a label saying: this was site A, channel webshop, project Zonnestraat. Three labels, one entry, one account.

The difference is that labels combine. You can look per site, or per channel, or per site within one channel, without having to create anything new. Add a site and you add one label, and your whole history stays comparable.

Meanwhile your chart of accounts stays what it should be: a tidy, limited list of accounts that follows your tax obligations.

What to watch for in practice

One dimension per question. Do not mix site and channel in one list, because then you are back at the multiplication. Separate dimensions for separate questions.

Keep the number of dimensions small. Two or three is workable. Set up six and you will get three that are filled in consistently and three that stay half empty, and the latter is worse than not having them.

Make them mandatory where you can. An analytical line that stays empty falls into the unknown category in your report, and as soon as that category gets big enough nobody trusts the report any more.

And fix the list with a date attached. If a label appears or changes meaning halfway through the year, you want to know from when on your series stops being comparable.

The part people underestimate

Working analytically feels like an accounting choice. It is not. It is a reporting choice you make inside your bookkeeping.

The question is not how to post correctly, because both approaches manage that. The question is which questions you want to be able to ask a year from now. Those questions determine your dimensions, and if you do not ask them before you start, you cannot answer them afterwards without relabeling your history.

That is a half day of conversation between the owner and the accountant. And it is one of the few moments where half a day earns you years of usable numbers.

What it does not solve

Two things I want to be honest about.

Analytical dimensions tell you nothing about what happened outside your bookkeeping. You can label revenue neatly per site and still not know how many people walked in.

And the reporting itself does not get better. In most packages, reading out analytical data is stiffer than you would like. It is an excellent way to structure your data, and usually still a reason to put your reporting somewhere else.

The claim

If your chart of accounts is growing because you want to see more, you are turning the wrong knob. Accounts are for the tax authorities. Dimensions are for you.