Manual tracking

Manual expense tracking, and what an automatic feed never sees

The gaps are not edge cases. They are the categories you most want to look at.

A manual expense tracker is one where you record each transaction yourself instead of importing it from a bank or card feed. It takes a few seconds per entry and it catches five things an automatic feed structurally cannot: cash, tips, splits, transfers between your own accounts, and refunds. It also means no app is holding a credential that reaches your money.

The case against manual tracking is obvious and usually correct: it is work, and work you have to keep doing. Most people who try it stop within a fortnight. So the honest version of this argument has to start by conceding that an automatic feed is easier, and then explain what the ease is buying and what it is costing.

What a feed can see, and what it cannot

An account feed sees movements through the accounts it is connected to. That is a good record of card spending and direct debits, and a poor record of almost everything else, because plenty of money moves without touching a connected account or touches it in a way the feed cannot interpret.

Cash

A withdrawal shows up as one line: two hundred at an ATM. Where that two hundred went — a market, three coffees, a haircut, a taxi — is invisible. Cash is a smaller share of spending than it used to be and it is still the least examined, precisely because it is the part no feed has ever explained. Tracking it manually is the only way it is ever tracked at all.

Tips

A card is authorised for the meal and settled later for the meal plus the tip. In between, an imported figure is wrong, and depending on when the feed read it, the figure it stored may stay wrong. Over a year of eating out, a tip is a meaningful line of its own that almost nobody has ever seen as one.

Splits

You put a hundred and twenty on your card and two friends send you forty each. The feed records a hundred and twenty of spending, and later, three transfers in, which it does not know are related. Your dinner category is now sixty pounds too high and your income is eighty pounds too high. Nothing is technically wrong and every number is useless.

Transfers

Moving money from a current account to a savings account is not spending. Paying a card bill from a current account is not spending either — the spending already happened when the card was used. Feeds routinely count both, which is how people end up looking at a month where they apparently spent more than they earned while their balances went up.

Refunds and reimbursements

A return that lands three weeks later is a credit with no visible connection to the purchase. Expenses your employer pays back are the same shape. Both quietly distort the category they belong to, in the direction that makes your record look worse than reality.

The credential question

To read your accounts, an app needs a way in. Depending on where you are that is a token issued through an open banking flow, or a set of credentials held by an aggregator, or in the worst arrangements your actual login, stored by a company you have never heard of because your budgeting app licenses their service.

None of that is necessarily reckless. It is, unavoidably, a database of routes into people’s accounts, sitting somewhere, worth a great deal to somebody. You are trusting an app you downloaded, a data aggregator you did not choose, and the security practices of both, forever, in exchange for not typing things in.

A manual tracker asks for none of it. There is no connection, so there is no credential, so there is nothing of that kind to lose. Whether that trade is worth a few seconds a day is a real question with a real answer on both sides — but it should be a decision rather than a default.

The reason that actually changes behaviour

If your only goal is a tidy archive of where money went, an automatic feed wins and this page is an argument for nothing. The reason to log it yourself is that it is the only version that changes what you spend.

An imported record tells you, three weeks later, that you spent four hundred on takeaways. You feel briefly bad and the number does not move. Recording it yourself puts a three-second pause between deciding and forgetting, roughly forty times a month. That pause is the entire mechanism. People who track manually and stick with it usually report that the first month felt like admin and the second month felt like the spending had quietly changed — not because anything told them to cut back, but because they had noticed.

This is also why a feed can feel oddly demotivating. It removes the only part of the process that was doing any work.

What makes people quit, and what to do about it

Almost nobody quits because they object to the idea. They quit because the app made each entry take thirty seconds: an amount field, a merchant field, a category dropdown, a date picker, a note, save. Forty times a month, that is twenty minutes of fiddly typing, and fiddly is worse than long.

The fix is to make entry as close to speaking as possible.

• Write a sentence, not a form. “18 on lunch” carries an amount, a merchant and a category already. Having to restate it in four boxes is the app failing to read something you already said.
• Log several at once. Most of the time you are catching up on the day, not recording one thing. One sentence should be able to produce four entries.
• Say it when your hands are full. Walking out of a shop is the moment you remember; it is also the moment you cannot type.
• Accept relative dates. “Friday” and “yesterday” are how people describe when something happened.
• Show it before saving it. Anything that parses a sentence will sometimes get it wrong, and a review step is what makes that a shrug rather than a corrupted month.

Do all of that and an entry costs about as long as unlocking your phone, which is the point at which the habit survives.

Where manual tracking is genuinely worse

It will not be complete if you stop doing it, and a partial record is worse than an honest gap because you may not remember which weeks you missed. It will not catch a subscription you forgot you had, which is the one job a feed does better than anything. And it will never reconcile to the penny against a statement, because you will round and you will forget the odd coffee.

If catching forgotten subscriptions is the whole reason you want a tracker, read your card statement once a quarter. That is a better tool for that specific job than either approach.

Common questions