← Blog Habit

Why you stop tracking expenses after a few days

A paper diary with a hidden sensor inside showed that only 11% of people filled it in as instructed. 90% said they had.

10 min read

Most people stop tracking expenses by hand during the first or second week, and the evidence points away from discipline as the cause. In 2002, when researchers used a hidden sensor to measure who was actually filling in a paper diary, 90% of participants believed they were complying and only 11% were. Manual tracking doesn't collapse for lack of will. It collapses because its cost lands at the worst possible moment and its payoff arrives weeks later.

The diary you think you're keeping isn't the one you're keeping

Arthur Stone, Saul Shiffman and three colleagues published a study in the BMJ with 80 chronic pain patients. Some were given a paper diary to record their pain three times a day; others got an electronic diary on a Palm. What the paper group didn't know is that their binder had a photosensor hidden in the spine, logging the exact time it was opened.

The paper group reported 90% compliance. The sensor said 11%.

Compliance they reported 90%
Compliance the sensor measured 11%
Stone et al., BMJ (2002): in the group keeping a paper diary, 9 in 10 said they were writing in it three times a day; the photosensor hidden in the binder found that barely 1 in 9 was.

They weren't lying. They were back-filling several days at once —Sunday night, before the appointment— and counting it, in good faith, as done. The diary existed. The data didn't.

That is exactly what happens to your expense tracking when you pick it up on Friday. You write down what you remember, and what you remember is round numbers, no merchant, no payment method. Precisely the detail that made the month legible.

Diary fatigue has had a name for thirty years

Statistical agencies have been fighting this problem for decades, with budget and method behind them. The US Consumer Expenditure Survey asks households to keep a two-week diary; in the 1987 round, spending reported in the second week ran 11% below the first. Those households didn't spend less. They wrote down less.

The same pattern is documented in Canada's Food Expenditure Survey and in the UK's Family Expenditure Survey. Methodologists gave it a name: diary fatigue.

It's worth pausing on who those people are. They formally agreed to take part, sometimes for an incentive. They know someone will read the notebook. They are writing for fourteen days in an instrument designed by professional statisticians. If reporting drops 11% in a week under those conditions, the app you downloaded on a Tuesday night never stood a chance.

App numbers, when measured rather than surveyed, are far worse. In 2019, Baumel and colleagues published panel-measured usage of dozens of mental health apps with more than ten thousand installs each. Median 15-day retention was 3.9%; at 30 days, 3.3%. These aren't finance apps, and that has to be said. But the mechanics are identical: log something every day in exchange for a benefit that shows up much later.

Why does expense tracking collapse when other things don't?

Logging an expense stacks three difficulties that are each hard on their own.

Motivation is a weaker lever than it looks

Thomas Webb and Paschal Sheeran reviewed, in Psychological Bulletin, the experiments that first changed someone's intention and then measured their behavior. The result: an intervention producing a medium-to-large change in intention (d = 0.66) produced only a small-to-medium change in behavior (d = 0.36). Convincing you to track your spending works. It works about half as well as it looks.

The habit takes longer than the attempt lasts

Phillippa Lally and her team followed 96 people for twelve weeks, each with a daily behavior of their own choosing, measuring day by day how automatic it was becoming. The median time to reach 95% of automaticity was 66 days, with a range from 18 to 254. Fewer than half the participants got there.

Sixty-six days. Diary fatigue eats the attempt in fourteen.

66 days nobody logs past here Day 1 7 14 30 66
Each bar is a day of logging; they go out on their own around the second week, which is where diary fatigue has been measured. The mark on the right is the 66-day median Lally et al. (2010) found for a daily behavior to become automatic.

Logging an expense means paying for it twice

This is the part almost nobody says out loud. Drazen Prelec and George Loewenstein described, in Marketing Science, what they called the coupling hypothesis: the tighter payment and consumption are coupled in time, the more paying hurts. It's why a running taxi meter ruins the ride, and why a flat rate feels cheap even when it costs more.

Now put that next to how you pay today. Cards, terminals and the iPhone exist in part because they decouple: you tap, it beeps, you feel nothing. Logging the expense couples it back. It's you, by hand and by choice, restoring the small sting the payment system had just removed.

The core of it: no finance app says this on its welcome screen, but it is what the app asks a person to do sixty-six days running. Preferring not to do it isn't a character flaw.

The cost isn't the seconds. It's the interruption

Gloria Mark, Daniela Gudith and Ulrich Klocke ran a CHI experiment on what happens when someone is interrupted mid-task. The finding was counterintuitive: the interrupted task got finished just as well, and faster. What went up was everything else. More stress, more frustration, more time pressure, more perceived effort.

Applied to the checkout line: a form isn't paid for in seconds, it's paid for in attention. And attention in line, with your cart blocking the aisle and someone waiting for you to put the phone away, costs far more than attention on your couch. That's why the expense you fail to log is almost always the same kind of expense: small, rushed, made while you were doing something else. Which is exactly the one you needed to see at the end of the month.

What can be designed, instead of what can be promised

If the problem is cost and timing, there are three levers, and none of them is willpower.

  • Remove decisions, not just steps. A typical form asks for amount, category, payment method, merchant and date: five decisions for a fact you already had complete in your head before opening the app.
  • Capture at the moment of the expense, not at the moment you remember it. It's the only way the data arrives with its detail intact.
  • Let the machine do the part nobody wants. Categorizing is admin work, and admin work is the first thing anyone drops.

In Sonanta we bet on those three. The budget is five seconds end to end: you say the sentence, it is interpreted inside the phone, and it is saved with a one-tap confirmation. That number is a design decision before it is a measured result. Past it, the hypothesis the whole thing hangs on says the habit falls apart.

With a form

  1. 1
    AmountType it in
  2. 2
    CategoryPick from a list
  3. 3
    Payment methodPick from another list
  4. 4
    MerchantWrite it out
  5. 5
    DateConfirm it
  6. 6
    SaveAnd back to what you were doing

With a sentence

I spent 800 pesos at Walmart on my Santander

Saved

Both paths end at the same saved expense. The left one asks for five decisions to get there; the right one asks for none, because the sentence already carried the amount, the merchant and the card.

What this evidence doesn't say

None of the above proves Sonanta works. It proves manual tracking fails consistently and measurably, which is a different claim.

There's also an easy explanation worth refusing: that "you run out of willpower." In 2016, a preregistered replication across 23 labs with 2,141 participants looked for the ego-depletion effect —the idea that self-control drains like a tank— and found d = 0.04, with a confidence interval that includes zero. Discipline failing to explain the drop-off doesn't mean there's a reservoir emptying inside you.

And the limits of what we're citing: Baumel's data is from mental health apps, not finance. Lally's study is 96 people with simple behaviors and self-reported automaticity. The BMJ work is chronic pain patients logging symptoms, not spending. None of the three was designed to talk about this. All three point the same way.

And ours: our own hypotheses —the five-second threshold included— are still marked unvalidated in the project documentation. We have a shipped app and a well-grounded bet, not a proof.

So the question left isn't how much discipline you're missing. It's how many days your current method would last if someone hid a sensor inside it.

Sources

Questions

Why do I stop tracking my expenses after a few days?
Because manual tracking charges its cost at the worst moment — in line, in a hurry, in the middle of something else — and delivers its payoff weeks later. Survey methodologists call it diary fatigue: in the 1987 Consumer Expenditure Survey, spending reported in the second week ran 11% below the first, among people who had formally agreed to take part.
How long does expense tracking take to become a habit?
Longer than the average attempt lasts. In Lally and colleagues' 2010 study, the median time for a daily behavior to reach 95% of its automaticity was 66 days, with a range from 18 to 254, and fewer than half the participants got there. An expense diary rarely survives two weeks.
Does expense tracking fail because of poor discipline?
The evidence does not support that reading. Webb and Sheeran (2006) found that a medium-to-large change in intention (d = 0.66) produces only a small-to-medium change in behavior (d = 0.36): motivation is a weak lever. And the idea that willpower drains like a fuel tank did not survive its multilab replication in 2016.
Is it worth logging expenses several days later?
Far less than it seems. In the 2002 BMJ study, paper-diary participants back-filled several days at once and still reported 90% compliance while the hidden sensor recorded 11%. An expense logged three days late arrives rounded off, with no merchant and no payment method: exactly the detail that made the month legible.