Why your spending categories keep drifting

Category systems don’t fail on day one. They fail on day ninety, when “Shopping” means five different things and you stop trusting the charts.

Drift is normal

Merchants change names. You invent a one-off category in a hurry because the row in front of you doesn’t fit anything. A trip creates “Travel” that never gets closed. A side project creates “Business” that overlaps three existing labels. Suddenly half your spend is “Other” and the rest is spread across near-duplicates.

Drift isn’t a character flaw. It’s entropy. Any taxonomy maintained by a human under time pressure will loosen, and the systems that survive are the ones designed to be re-tightened rather than the ones designed to be perfect.

Symptoms worth acting on

• More than roughly 15% of spend sitting in catch-alls
• Duplicate near-synonyms (Food / Groceries / Dining / Restaurants)
• Categories you haven’t used in six months but still scroll past
• Arguments with a partner about what a label means
• A month where you tagged the same merchant two different ways
• Charts you look at, distrust, and then ignore

The last one is the real signal. A taxonomy you don’t believe has already stopped doing its job, whatever the numbers say.

Why it happens faster than you expect

Three forces do most of the damage. Ambiguous boundaries, where a category could plausibly hold a transaction and so sometimes does. Too many options, because a picker with forty entries guarantees inconsistency; nobody scrolls that far twice. And unreadable rows, where you can’t tell what a charge was and default to a catch-all.

That third one is why clean merchant names matter more than a clever tree. Most “Other” isn’t indecision. It’s illegibility.

A lightweight reset

Once a quarter, spend twenty minutes:

Sort the last three months by category and look at the top of the list and the bottom. Merge synonyms, keeping the name you’d say out loud. Kill orphans with one or two transactions by folding them into a parent. Then rewrite three rules you could explain to a stranger, such as “coffee out is Dining, coffee beans are Groceries,” and write them down somewhere in the workbook.

Written rules are what stops the same argument from recurring in February. An unwritten convention is not a convention.

Don’t rebuild from zero unless the tree is genuinely unusable. Full resets feel productive and get abandoned around week three, and they break every historical comparison you had. Attach the reset to your monthly money review every third month, so it’s a scheduled chore rather than a burst of enthusiasm.

Rules of a maintainable tree

Fewer categories than your ego wants. Ten to fifteen is plenty for most households; the urge toward forty is a desire for precision that the data won’t reward.

Separate only when the decision changes: needs vs wants, fixed vs flexible, tax-relevant vs not. If a category never changes behavior, it doesn’t need to exist. “Streaming” split from “Subscriptions” earns its place only if you’d actually cancel differently.

Sheetful keeps feeding transactions into the sheet, automatically each day on Premium or manually whenever you choose on Free. You still own the taxonomy, and that’s the right split: the feed is mechanical, the meaning is yours.

A boring tree you maintain beats a perfect tree you abandon.