Why Purchases Under $15 Feel Free: The Research, and a One-Month Log to Check Your Own
Small purchases slip past the budget because of how they are paid, not how much they cost. The payment research, plus a 30-day log that needs no app.
Say you open a bank statement and find six or seven charges you cannot place. None is large. Each is under fifteen dollars. You could name the merchant on maybe half of them and the purchase on none. That is a common enough experience that the question is worth asking properly: why does a class of purchase exist that leaves no memory at all, and what would it take to see it?
This piece has no log of mine in it and no total. It sets out what the payment and mental-accounting research says about why small purchases behave differently from large ones, then gives a thirty-day log method you can run on your own statement, and finally what to do with the result. Where there is a number, it comes from a published source with a link and a date.
Small purchases are most of your purchases
Start with how many there are. The Federal Reserve’s 2026 Diary of Consumer Payment Choice, which asks a panel of US consumers to record every payment for three days, found the average adult made 47 payments a month in 2025, and that 17 of those were in-person payments under $25 (Federal Reserve Financial Services, 2026 findings, PDF page 11, checked September 18, 2026). Roughly a third of everything you pay for, by count, is a small in-person transaction. Add remote ones (app stores, deliveries, one-off online purchases) and the share by count is higher still.
The same report describes what changed. The number of under-$25 payments has been steady since 2017, but the number paid in cash has fallen by nearly half over that period. Consumers moved small purchases onto cards and phones and did not move them back after the pandemic. Cash is now mostly a small-purchase instrument, five out of six cash payments are under $25, but it is a shrinking one.
That matters because the payment method, not the amount, is what the research says governs whether a purchase registers.
What the research says about why they do not register
| Finding | Source | What it means for a $9 purchase |
|---|---|---|
| Paying feels like a loss at the moment of payment, and the strength of that feeling depends on how coupled the payment is to the consumption. Cash is tightly coupled; cards and prepaid balances decouple the two. | Prelec & Loewenstein, “The Red and the Black”, Marketing Science 17(1), 1998 (DOI) | A tap that is settled later against a balance you never look at produces almost none of the “pain of paying” a $9 note would. |
| People run separate mental accounts and evaluate outcomes account by account, not against total wealth; small amounts fall below the threshold at which an account gets opened at all. | Thaler, “Mental accounting matters”, Journal of Behavioral Decision Making 12(3), 1999 (DOI) | Each $9 is judged on its own, where it is trivially affordable; the sum of forty of them is never judged at all. |
| Asked to estimate monthly subscription spending and then to itemise it, 1,000 US consumers estimated $86 on average and itemised $219; 74% said recurring charges are easy to forget. | C+R Research 2022 survey (summary, checked September 17, 2026) | The category people are worst at estimating is precisely the one made of small automatic charges. One commercial survey, but the direction is consistent with the two papers above. |
| Cash payments concentrate in small transactions; card use replaced cash for small purchases after 2020 and stayed there. | Fed Diary of Consumer Payment Choice, 2026 findings (above) | The purchases most likely to be invisible are increasingly paid in the way most likely to make them invisible. |
Put together: a purchase under some threshold, paid without a visible transfer, is evaluated alone, feels close to free, and is forgotten. None of these papers gives a universal threshold. Fifteen dollars is a round number I use in the method below because it sits above a coffee and below almost anything you would call a decision; if your own line is ten or twenty, use that.
The one-month log
The point of the log is to make the pile visible once, not to run it forever. It works without an app, which matters because a tracking habit that needs setup is one most people drop within days.
- Pick the threshold and write it down. Everything under it, any payment method, any channel, goes on the log. Everything over it does not. The over-threshold purchases are already visible to you; that is the whole premise.
- One line per purchase, at the moment of purchase. Thing, amount, day. Use the notes app already on your phone. Do not categorise in the moment; capture only. If you miss one, add it from the statement that evening rather than reconstructing it a week later.
- Include renewals. Small app subscriptions and in-app purchases usually appear on the statement, not at a till, and they are the ones the C+R survey suggests people underestimate most. Once a week, check the statement for anything under the threshold you did not log by hand.
- Do not change behaviour during the month. If the log makes you buy less, you have measured the log, not your spending. Decide in advance that the first month is observation only.
- At thirty days, sort three ways. By amount, to see the total. By channel (till, app store, delivery, transport, renewal), to see where the invisible ones came from. By whether you can recall the purchase without the note, which is the column that tells you which purchases were decisions and which were reflexes.
Reading the result
I will not tell you what your total will be, because nothing in the research predicts it, and a headline figure from one person’s month would be exactly the kind of number this site has stopped publishing. What the research does predict is the shape of the result.
Expect the digital column to surprise you more than the physical one. A coffee leaves a cup in your hand; a $7 in-app purchase leaves nothing, and Prelec and Loewenstein’s decoupling argument says that is the one you will not remember. Expect renewals to be the largest single line in the “cannot recall” column; that is the C+R gap showing up on your own statement. And expect the total to be dominated by count, not by any individual line, which is Thaler’s point: no single entry was worth a mental account, so none of them got one.
Two things the log is not. It is not evidence that the purchases were wrong. Some share of them will be things you would buy again with full attention, and the exercise is meant to find that share, not to shame it. And it is not a budget; it produces a single month’s view, and a single month is not a trend. Run it twice, a few months apart, before drawing any conclusion about direction.
What to do with the pile
The change I would make, having looked at a month of small purchases, is not a cap. It is a delay, applied only to the reflex column.
A 24-hour rule for non-essential small purchases. Anything under the threshold that is not food or a renewal you have already decided to keep gets written down with no action. If you still want it the next day, buy it. The reasoning follows from the decoupling research rather than from willpower: the purchase felt free because the payment was instant and invisible, and a day’s delay reinstates the decision that the tap removed. How many survive the day is something you will find out; I am not going to claim a proportion.
A renewal review, separately. The recurring lines want a different treatment from the till lines, because they repeat without a new decision each time. Small subscriptions are the hardest to cancel for the same reason they are hard to notice, and cost per use is a better test for them than “is $6 a month a lot”.
Leave the coffee alone, or do not, but decide once. The goal is not to drive the number to zero. It is to move the purchases that were made because the moment was there, not because you wanted them, back into the category of things you decide. If a small purchase passes the 24-hour test every time, take it off the log and stop spending attention on it.
A month of this produces one thing of value: a list of purchases you cannot explain, which is a more useful input than any category total. The unexplained-purchase bucket is the method for what to do with that list; tracking every purchase for thirty days is the same log without the threshold, if you would rather see the whole statement at once.
Sources checked September 18, 2026: the Federal Reserve Financial Services 2026 Diary of Consumer Payment Choice report (linked above; the under-$25 figures are on page 11 of the PDF), the Prelec & Loewenstein and Thaler abstracts via the DOIs above (both resolve to the publisher pages; the one-line characterisations in the table are mine), and C+R Research’s published survey summary. This article is general information about tracking your own spending, not financial advice.
Related reading
The Budget Category No App Offers: "I Don't Know Why I Bought This"
Add one bucket for purchases you cannot explain, then sort it by what went wrong, not what was bought. The method, plus the research behind it.
Does Paying in Cash Make You Spend Less? What the Research Says, and the Part You Can Keep
Studies do find people pay more with cards than with cash. What they measured, what they did not, and a two-minute habit that keeps the useful part.
How to Log Every Purchase for 30 Days, and What the Log Is Likely to Show
A one-month purchase log is the cheapest spending audit there is. A method that survives the month, plus sourced benchmarks to read your results against.
Why a Subscription You Decided to Cancel Keeps Renewing
Status quo bias, billing inattention and a missing trigger keep a decided cancellation alive. A dated if-then fix, plus where click-to-cancel law stands.