How do I start tracking expenses if I have never done it before?

Do not start by importing history or building categories. Log the very next thing you buy, in whatever way takes the least effort, and do that for seven days with a rough category on each entry. After a week you have real data and you know whether your capture method is fast enough to survive. Set budgets only after that.

Should I track expenses manually or use automatic bank sync?

They do different jobs. Bank sync is good at completeness and poor at meaning: it captures everything but labels it with merchant strings, and it needs your banking credentials. Manual capture is better at awareness, because the moment you log a purchase is itself the moment of reflection. If you want a tidy record, sync. If you want to spend less, capture manually - as long as it is fast.

How do I track cash spending?

Pick one rule and keep it. Either log the ATM withdrawal once and treat that money as spent, or log each cash purchase individually and treat the withdrawal as a transfer. Both work. Mixing them produces totals that do not reconcile and a system you stop trusting.

What should I do if I miss a week of tracking?

Restart today and do not reconstruct the gap. Rebuilding a missed week from memory takes an hour and produces bad data. Mark the month as incomplete and resume with your next purchase - the value is in the trend across months, and one incomplete month barely disturbs it.

How do I track expenses in more than one currency?

Record each expense in the currency you actually paid in and let the app convert for reporting. Converting in your head at the point of entry embeds a different wrong rate into every entry and makes month-to-month comparison meaningless.

Blog

The Expense Tracker Guide That Survives Week Three

Updated August 2026

Almost nobody quits tracking expenses because they lack discipline. They quit because logging a coffee takes forty seconds and eleven taps. This guide is built the other way round: capture first, categories second, budgets last.

Tracking is an entry problem, not a willpower problem

The usual advice is to categorise carefully, review regularly and stay consistent. It is not wrong, it is aimed at the wrong bottleneck. Nobody stops tracking during the monthly review. They stop somewhere between the till and the car park, three seconds after deciding it can wait.

One deferred entry becomes four, four becomes a backlog, and by week three the app is a red badge you swipe past. The fix is not more resolve - it is making entry cheap enough that deferring it is not worth the trouble. The consequence is unusual: most setup work - category trees, budget limits, tags - is worth doing after your first month of data. Setup done in advance is a guess.

Step 1: Decide which question you are tracking to answer

"Tracking my expenses" is not a goal, and treating it as one is why so many setups collapse under their own detail. There are three different jobs here, each justifying a different amount of effort.

Awareness - you want the shape of a normal month. Broad categories are plenty and precision to the cent is pointless. Most people start here and never need more.
Control - you already know the problem area. Detail earns its cost, but only inside the one or two categories you are working on. Leave everything else coarse.
A shared household - what matters is not granularity but visibility: both people entering into the same place, both seeing the same live total, without anyone asking for a screenshot.

Most guides assume you are in the second group. Most readers are in the first. Building a control-grade system when you only need awareness is the most common way to over-engineer yourself out of the habit.

Step 2: Choose a capture method and time it with a stopwatch

This one decision determines whether you are still tracking in three months, so measure it. Log a real purchase with your candidate method, standing up, hands full. Under ten seconds survives a checkout queue. Over thirty seconds does not survive a bad mood.

Method	Time per entry	In practice
Say it out loud	About 3 seconds	Works while walking or carrying bags. No screen, no category picker.
Photograph the receipt	About 5 seconds	Fastest at the till, but only half an entry until something parses it.
Type one line	About 10 seconds	Fine at a desk. Slows down badly one-handed or in a queue.
Fill a structured form	About 30 seconds	Amount, date, category, account, note. Accurate, and the first thing people abandon.
Save receipts, log later	20 minutes weekly	Batching moves the cost into one dreaded block. Skipped weeks compound.

Speaking wins for a structural reason, not a novelty one: it is the only method that does not compete with whatever your hands are already doing. The caveat is that voice only works when something parses it - a voice note you transcribe later is worse than typing. What makes it viable is the amount, category and date being extracted for you, so the spoken sentence is the entry. That is the whole point of a voice expense tracker.

Step 3: Start with seven categories, and earn the eighth

Categories are not a filing system. They are questions you are prepared to answer later, and a category you never look at is pure entry cost. A set that covers almost everyone: groceries, eating out, transport, bills and subscriptions, shopping, health, and everything else.

Two rules keep it working. First, the two-second rule: if an entry does not obviously belong somewhere within two seconds, it goes in "everything else". A slightly wrong category is worth far more than an unlogged expense. Second, split a category only when it has failed you - when you genuinely wanted to know something it could not tell you. That is the difference between a category you need and one that looked tidy in a template.

Automatic categorisation changes the maths. If the app assigns a category from what you said or photographed, categories cost nothing at entry and you can afford more of them - though you still only benefit from the ones you read.

Step 4: Fix one rule each for cash, income and shared spending

These three are where systems actually die - not in dramatic abandonment, but in a slow loss of trust as the totals stop matching reality. Each needs exactly one rule, decided once.

Cash - either log the ATM withdrawal once and treat that money as spent, or log every cash purchase and treat the withdrawal as a transfer. Never both. A hybrid produces double-counted months.
Income - track it separately, from the day it lands. Expenses alone tell you where money went, never whether it was affordable. This matters most for irregular earners, where the useful planning number is your lower typical month, not your average one.
Shared spending - household costs belong somewhere both people can add to and both can see; personal spending stays personal. Merging the two turns into mutual auditing or into nobody logging anything. The separation is what makes shared tracking last, and it is exactly what a spreadsheet is worst at.
Foreign currency - store the amount you actually paid, not your mental conversion of it. Converting at entry embeds a different wrong rate into every entry and makes month-to-month comparison meaningless.

Step 5: Set budgets from your data, not your aspirations

Do this in month two. A limit invented before you had data is a wish, and a wish you breach on the 14th every month teaches you to ignore the app.

Start at your actual average, not below it. That sounds pointless - you already spend that - but it converts an invisible number into a live progress bar, and seeing "70% with ten days left" changes decisions without imposing any restriction. Once a limit holds, cut it by ten percent, one category at a time. Cut five at once and you will breach all five and conclude that budgeting does not work for you.

Leave fixed costs unbudgeted - rent and insurance do not respond to a progress bar. Budget only the categories where a mid-month decision is still available to you.

Step 6: Three minutes weekly, twenty monthly, one change

The value is in the entry, not the review. Reviews only need to be frequent enough to catch problems while you can still act on them, and rare enough that they never feel like admin.

Weekly, three minutes: glance at category totals looking for one thing only - anything near its limit with most of the month left. Monthly, twenty minutes: compare against last month, fix any limit that was consistently wrong, and note the irregular costs coming up. Most "unexpected" expenses were entirely predictable a month out.

Then change exactly one thing. A single change you keep beats four you abandon, and you learn which one mattered. The most reliable first change is rarely a spending category - it is a subscription audit. Filter to recurring charges and read the list out loud.

When you fall off, restart at today

You will miss days. The only thing that decides whether the habit survives is what you do on the day you notice. Do not reconstruct: rebuilding a missed fortnight takes an hour, produces data you know is partly invented, and converts a small lapse into a chore large enough to avoid. Mark the month incomplete, log the next thing you buy, carry on.

If you fall off repeatedly at the same point, that is diagnostic rather than moral. Repeated gaps almost always mean your capture method is too slow for the situation you keep failing in - the commute, the supermarket, the evening out. Go back to Step 2 and make entry cheaper for that specific case.

How Peggy handles this

Peggy was designed from the entry-cost problem backwards. Say "twelve on lunch" and it parses the amount, picks the category and writes the entry - no form, no picker. Categories cost nothing because they are assigned for you. Shared household costs live in a space both people add to and both see live, invited with an 8-digit code rather than bank credentials, while personal spending stays out of it. Every transaction stores the currency you actually paid in. And adding an expense works offline, on iOS and Android, because the moment you need it most is the moment you have no signal.

Comparing tools rather than methods? See Peggy against YNAB, Mint and Monarch, or read the AI budget app guide.