Trixie
← All articles

Financial planning · August 24, 2026 · 6 min read

Your monthly average is lying to you

Here's a question your budgeting app probably answers wrong: how much do you spend per month? Not last month specifically — in general. The number you'd use to decide whether you can afford a bigger apartment, how much you can invest, or how big your emergency fund should be.

The naive answer is to average your last few months of spending. And the naive answer breaks the moment real life happens. Buy a $40,000 car in March and your "average monthly spending" jumps by thousands of dollars for the rest of the year. Every downstream number — your savings rate, your emergency fund target, your retirement projection — inherits the distortion.

Two kinds of spending

The fix is to admit that spending comes in two kinds. Run-rate spending is the recurring cost of your life: housing, food, subscriptions, gas, the things that happen every month whether you think about them or not. One-off spending is exceptional by nature: the car, the wedding, the emergency vet bill, the once-a-decade roof repair.

Both are real money and both belong in your total. But they answer different questions. Run-rate answers "what does my life cost?" One-offs answer "what did this year cost?" When an app mixes them into a single average, it answers neither.

What honest metrics look like

The transfer problem

There's a second, sneakier distortion: internal transfers. Move $5,000 from checking to savings and many apps count it as $5,000 of "spending" — then count the arrival as $5,000 of "income." Pay your credit card bill and the payment double-counts with the purchases it covered. If your app's cash flow numbers have ever felt vaguely inflated, this is usually why.

Cleanly separating real cash flow from internal movement — transfers between your own accounts, card payments, loan mechanics — is tedious to do by hand, which is exactly why software should do it. When it does, your monthly numbers get boring. Boring is what accurate looks like.