SpendTrackSpendTrack

Where Your Money Goes: How to Rebuild Your Spending Map from the Last 3 Months in One Afternoon (Without Connecting Your Bank)

Team at SpendTrackWritten with AI assistance19 min read

In summary

  • Manually tracking expenses almost never lasts. The shortcut that does work is the opposite: reconstruct the last 2-3 months from the statement that your bank already allows you to export, and only then keep it up to date.
  • According to the INE Family Budget Survey (2025 data, published June 25, 2026), the average Spanish household spent €35,101 in the year. Housing and food account for 49.2%; the category where there are indeed decisions (transport, dining, and leisure) is around 25.8%: about €755 per month that almost no one can break down from memory.
  • There are two pools of money that don’t appear in your head but do in the statement: subscriptions you no longer use and small, repeated, invisible expenses.
  • SpendTrack, the mini expense tracking app from AIsToHire, is designed specifically for that method: it reads the statement in CSV, TXT, or PDF, suggests a categorization for each transaction, and does not save anything until you confirm. It never asks for access to your bank account.
  • This is a method for household organization, not financial or tax advice. SpendTrack records and summarizes what you input; it does not move money, does not hire anything, and does not replace a professional.

If you’ve arrived here by searching on Google "where does my money go", the problem is not that you spend poorly: it’s that you don’t have the data. You earn, you pay, and at the end of the month, a chunk is missing that you can’t explain. This article is the concrete method to extract that data in one afternoon, step by step, and maintain it afterwards with almost zero effort. We’ve written it around SpendTrack, the personal expense tracking app from AIsToHire, because it’s the tool we use — but the five steps are equally valid with a spreadsheet, only it will take you quite a bit longer.

Why you don’t know where your money goes (and it’s not due to disorganization)

You don’t know because your memory retains the big expenses and erases the small and repeated ones, which is exactly where the leak is. You remember the rent; the 23 coffees, the four takeout orders, and the subscription you renewed in March, you do not. And those three things together weigh more than the dinner you do remember.

The INE numbers explain it well. These are the major spending groups of the average Spanish household in 2025:

Spending Group€ per year per household% of budgetVariation 2024→2025
Housing, water, electricity, gas€11,66533.2%+5.8%
Food and non-alcoholic beverages€5,62616.0%+4.4%
Transport€4,02011.5%+3.7%
Restaurants and accommodation€3,2829.3%−2.7%
Leisure, sports, and culture€1,7725.0%+4.7%
Health€1,4134.0%+2.6%
Total household€35,101100%+3.1%

Source: INE, Family Budget Survey 2025, final results published June 25, 2026.

Notice the distribution: half of the budget (49.2%) is housing and food, which you can hardly move from one month to another. But transport, dining, and leisure add up to 25.8%, about €9,056 per year, €755 per month. That’s where decisions need to be made — and it’s exactly the part that no one can reconstruct from memory.

Note: those figures are the national average of households of all sizes, not "what you should be spending". They serve to understand the shape of the problem, not as a target.

The quick way to know is not to track: it’s to reconstruct

Starting to track expenses today means having your first useful data in a month, and that’s if you last through four weeks. Reconstructing backwards from the statement gives you three months of data this very afternoon. It’s the difference between a promise and a diagnosis.

The method has three phases and does not change regardless of the tool you use: extract the history → classify it → separate it into fixed and variable.

Three panels on a wooden table: bank statement sheets coming out of a folder, hands sorting tickets into four piles with color marks and six piles of coins of different heights ordered from tallest to shortest
Three panels on a wooden table: bank statement sheets coming out of a folder, hands sorting tickets into four piles with color marks and six piles of coins of different heights ordered from tallest to shortest

How long it really takes, with the statement already downloaded: between 20 and 40 minutes for three months of a normal account. What takes time is not classifying, it’s reviewing: looking line by line and recognizing what each charge was.

Step 1: get the statement from the last 3 months from your bank

Log into online banking from your computer (not from the mobile app, which almost never allows exporting) and look for the option Export / Download transactions in the account details. Request the range of the last 90 days and choose CSV if it's offered; if not, the PDF of the monthly statement will suffice.

What format to choose, in order of preference:

FormatResult qualityWhen to use it
CSVThe best: clean columns, exact dates and amountsWhenever the bank offers it
TXT / CSV with unusual separatorsGood: you need to detect the decimal separatorOld banks and "flat" exports
PDF of the statementAcceptable: text is extracted, may lose some linesWhen there is no option to export data
ScreenshotNot useful—

In SpendTrack, that file is uploaded from + → Statement. Real limits of the app, so you’re not caught by surprise: 5 MB per file, up to 60,000 characters of content analyzed per pass, and 1,000 transactions per confirmation. For three months of a household account, you’re good; if you have a whole year, split the statement by quarters.

The important thing about this step: the file is not saved. SpendTrack reads it, returns the list of normalized transactions on screen, and only persists those you confirm. And at no point is access to your bank requested: no credentials, no aggregation, no connection to revoke later.

Step 2: convert 300 lines of statement into 14 boxes

A statement is useless as is: "PURCHASE CARD 4567 MERCADONA 2143 VALENCIA" tells you nothing aggregated. The real work is putting each line into a box and then summing boxes, not lines.

SpendTrack uses 14 closed typologies: housing, utilities, food, transport, dining, leisure, health, subscriptions, education, insurance, taxes, clothing, gifts, and others. You can add up to 10 of your own. And they are few on purpose: with 40 categories, you spend half an hour deciding if a toll is "transport" or "travel," and the summary becomes unreadable.

When importing the statement, each transaction arrives with a proposed typology, and you correct it on the review screen before confirming. That correction is not lost: the app's classifier works on three levels, in this order:

  1. Your fixed rules. If you create the rule "if the description contains AMZN → Leisure", it sends that, always. And you can apply it to expenses you already had recorded.
  2. Merchant memory. When you correct the same merchant twice, the app remembers it and from then on classifies it on its own.
  3. The AI proposal. Only when there is no rule or memory. If confidence is low, the expense is marked as "review" with an amber point instead of silently slipping into the wrong category.

Practical advice for this step: don’t chase 100%. If 90% of the transactions fall into the correct box, the map is already useful. The remaining 10% is fixed by the merchant memory alone in the second import.

Step 3: separate fixed from variable before judging anything

This is the step that almost everyone skips, and it’s the one that turns a list into a decision. A monthly total means nothing if you don’t know what part is inevitable. Spending €1,800 with €1,300 in fixed costs is a completely different situation from spending €1,800 with €700 in fixed costs.

Fixed expenseVariable expense
What it isSame amount and same date every monthChanges depending on what you do
ExamplesRent or mortgage, utilities, insurance, taxesShopping, dining, leisure, clothing, transport
What you can do this monthAlmost nothing: need to renegotiate or change providerA lot: this is where a decision shows in days
Why it’s useful to knowIt’s your floor: you can’t go below itIt’s your real margin of maneuver

SpendTrack marks housing, utilities, insurance, and taxes as fixed by default, and you can change that list in the settings. Why does it matter? Because the app uses that separation for two very specific things: to calculate what you really have available (subtracting the fixed costs that haven’t hit this month yet) and to project the month-end without contaminating the rhythm with an annual bill that just came in. If you touch the total of the summary, you’ll see the breakdown of fixed and variable without having to calculate anything.

Step 4: subscriptions are the easiest money to recover

Of all the things you will find in the statement, forgotten subscriptions are the only expense you can eliminate without changing your life at all. You are not giving up dinners or trips: you are stopping the payment for something you do not use.

The size of the problem, with source: in the Self Financial study from March 2026 on 1,272 adults in the United States, 59.9% admitted to paying for at least one subscription they do not use, with an average of 2.6 dead subscriptions and $26.79 a month wasted — out of $35.03 total spending on subscriptions. That is, three out of every four subscription dollars were going to dormant services. These are data from the U.S. and cannot be directly transferred to Spain, but the pattern — forgetfulness, not waste — is the same everywhere.

Five identical coins placed at equal distances on a straight green thread, with loose coins and papers scattered around without a pattern and a magnifying glass over one of the repeated coins
Five identical coins placed at equal distances on a straight green thread, with loose coins and papers scattered around without a pattern and a magnifying glass over one of the repeated coins

The good thing is that a subscription reveals itself in the statement: it is the same merchant, charging almost the same amount, at regular intervals. That is exactly what the SpendTrack detector looks for, and these are its real rules:

ConditionThreshold that the app applies
Times the same merchant appearsAt least 2 charges
Stability of the amountWithin the median ±10%
Average interval between charges (weekly)Between 5 and 9 days
Average interval between charges (monthly)Between 25 and 36 days
Average interval between charges (annually)Between 330 and 400 days
Estimated next chargeLast charge + detected period

That’s why importing three months at once is so profitable: with just one month, the detector has nothing to compare; with three, all the monthly ones come out. The detected ones appear in a list with their next charge, the app alerts you before it arrives (the days in advance are configurable) and you can mark one as ignored if it is a recurring charge you want to keep.

The ones you want to cancel, you cancel yourself in the corresponding service: SpendTrack does not cancel anything or manage cancellations. What it does is tell you which ones you have, how much they total, and when they will charge you again.

Step 5: the small, repeated expenses are not in your memory, they are in the pattern

A small expense is not a small expense: it is a small expense that repeats. That’s why you do not detect it by reviewing your list —each line seems irrelevant— and a count does detect it.

SpendTrack marks them with an explicit criterion: expenses of less than 5 in your base currency that repeat 5 times or more in the same month at the same merchant. A coffee costing €1.60 is nothing; 22 coffees are €35 a month and €420 a year. The point is not to stop drinking coffee: it is that the figure exists so you can decide with it.

In the Analysis view, alongside the small, repeated expenses, you will also find the trend of the last 12 months, the breakdown by tags (useful for things that cross categories, like "Trip to Lisbon") and by payment method, the peaks of the month, and seasonality. These are the four angles that explain almost any unusual month.

Step 6: set a number for the month and have it alert you before you overspend

A spending map without a budget is archaeology. The budget is what turns the historical data into a signal that arrives on time, while you can still do something about it.

In SpendTrack, you set a global budget and, if you want, one by type, and the app alerts at 80% and 100% of each. These are the calculations it makes, so you know exactly what you are looking at:

IndicatorHow it is calculatedWhat it is for
AvailableBudget − spent this month − fixed costs still to comeWhat you really have left, not what it seems
Available per dayAvailable ÷ days left in the monthYour sustainable pace from today
Closing forecastSpent + (daily variable pace × days remaining) + pending fixed costsWhere you end up if you keep going the same way
Day you overspendCurrent day + (budget − spent) ÷ daily paceThe alert with margin to react

These are deliberately simple formulas, and the app shows them as they are, without black boxes. If the month’s pace is atypical —a trip, a move— the forecast exaggerates it, and that is useful information too.

Two honest warnings about this: the first is that SpendTrack is not an advisory tool; it does not tell you where to invest, does not assess whether your budget is reasonable, nor does it replace an advisor. The second is that notifications require having the app open: there are no push alerts with the application closed, because it is a mini web app. The alerts accumulate in a notification center and you see them upon entering.

How to keep the map alive without dedicating time

This is where expense control systems fail: the history is done, but day-to-day requires typing. The solution is not more discipline, it is to have five entry points and use the one with the least friction at that moment.

A mobile phone in the center of a wooden table connected by green threads to five objects: a crumpled receipt, a microphone, a numeric keypad, a paper speech bubble, and a folder with sheets from the statement
A mobile phone in the center of a wooden table connected by green threads to five objects: a crumpled receipt, a microphone, a numeric keypad, a paper speech bubble, and a folder with sheets from the statement
Entry pointWhen is it bestWhat you have to do
Quick entry (2 taps)Cash payment, standing, in a hurryType the amount and touch the type. That's it
Photo of the receiptPurchase with several items or a receipt you want to see laterTake the photo and continue with your business: it is analyzed in the background
Voice dictationYou are walking or driving and just paidLong press on the + button and say it out loud
ChatYou remember an expense from yesterdayYou write "coffee 3.50 yesterday" to the assistant
Bank statementMonthly update of what was paid with a cardUpload the file and confirm what is missing

Three details that matter more than they seem:

  • The photo of the receipt does not block. It uploads, queues, and the AI reads it while you continue using the app; the result waits for you in a tray until you review it. Accepts JPG, PNG, WebP, HEIC, and PDF up to 15 MB, and if there are several receipts in one photo, it separates them into several expenses.
  • The dictation does not upload your audio. The transcription is done by your mobile's browser with the Web Speech API; what travels to the server is the already transcribed text.
  • Before creating anything, it warns if it was already there. By identical file, by invoice number with the same amount, and in the statements by a fingerprint of date + amount + description. This allows you to import two overlapping months without duplicating half a dozen receipts.

What SpendTrack DOES NOT do (and it's good to know before starting)

We say this here so you don't discover a limit halfway through. SpendTrack is a personal expense record, not a bank aggregator or a financial service.

Does not doWhat is in its place
Connect to your bank or read your transactions liveManual import of the statement that you export (CSV/TXT/PDF)
Cancel subscriptions or manage cancellationsDetect them, notify you of the next charge, and sum them up
Provide financial, tax, or investment adviceA summary of your own data, for you to decide
Notify you with the app closed (push)Notification center and alerts with the app open
Shared account, family, or with permissionsShared expenses with splits and pending balances, within your account
Track investments or mileage—

And what it does guarantee, because it's part of the design: your data comes out whenever you want in CSV, in a PDF report for the period you choose, and in a complete backup that can be re-imported. What you delete goes through a 30-day trash before disappearing, and from Settings, you can delete all your data, including receipt files. Each expense retains its original currency (45 currencies), the counter value, and the exchange rate on the day of the expense, so a trip doesn't throw off your month.

The three ways to know where your money is going, compared

There is not just one way to do this. These are the three real ones, with what each costs:

Notebook or spreadsheetApp connected to your bankRebuild from the statement (SpendTrack)
First useful dataIn 1 monthIn minutesIn an afternoon
You have to provide bank credentialsNoYes, to an aggregatorNo
ClassificationManual, 100%Automatic, little correctableProposed and reviewed by you before saving
Cash and non-card expensesYes, if you note themUsually does not see themYes: photo, voice, chat, or 2 taps
Maintenance effortHigh and constantVery lowLow: only what doesn't go through a card
Where it failsAbandoned in 2-3 weeksConnections drop and categories don't matchYou have to remember to import the statement

The third column is a deliberate middle ground: you forgo total automation in exchange for not handing over the keys to your bank to anyone and for nothing to enter your accounts without you having seen it first.

Frequently Asked Questions

How do I know where my money goes if I pay almost everything with a card?

Export the statement from the last 90 days from your online banking on your computer and classify it by categories. With a card, everything is recorded: the work is just to group it. In SpendTrack, you upload that file (CSV, TXT, or PDF), review the transactions with the category that the app suggests, and confirm. In 20-40 minutes, you have three months mapped out.

Do I have to give my bank credentials or connect my account?

No, and in SpendTrack, you can't even if you wanted to: the app does not request access to bank accounts nor does it use aggregators. You export the transaction file from your bank and upload it. That file is not stored: only the expenses you confirm on the review screen are saved.

How many months back is worth importing?

Three is the minimum useful and the best balance point. With just one month, you can't distinguish a one-time expense from a recurring one, and the monthly subscription detector needs at least two charges from the same merchant to identify the pattern. More than six months adds review work and little new information.

And what about cash expenses that don't show up on the statement?

You have to enter those yourself, but without typing forms: quick entry in two taps (amount + category), photo of the receipt, voice dictation, or a message to the assistant like "coffee 3.50". The cash route is precisely where the apps connected to the bank go blind.

How much time do I really have to dedicate each month?

After the initial setup, the actual maintenance is a couple of loose minutes a day (cash expenses as they occur) plus one import of the statement per month. What takes time is reviewing, not classifying: the app suggests the category, and when you correct the same merchant twice, it stops asking you about it.

Does AI make mistakes reading receipts and statements?

Yes, it can make mistakes, and that's why the flow is designed for you to see it first. In the statement, nothing is saved until you confirm; in the receipts, the result stays in a tray waiting for your review, and when confidence is low, the expense is marked as "review". Amounts and dates are always correctable.

Does this serve as accounting or for tax declaration?

No. SpendTrack is a personal expense control tool, not an accounting or invoicing program, and it does not provide tax advice. You can export your data to CSV or a PDF report for your use or to give to your advisor, but accounting or tax obligations are fulfilled with the appropriate tools and professionals.

What happens to my data if I stop using the app?

You take it with you: export to CSV, PDF report for the period you choose, and a complete backup that can be re-imported. And if you want to leave completely, from Settings, you can delete all your data, including the receipt files. What you delete goes through a 30-day recycle bin in case you change your mind.

Does it work if I spend in multiple currencies?

Yes. Each expense retains its original currency —there are 45 available—, the equivalent in your base currency, and the exchange rate applied on the date of the expense, not today. Totals and budgets are always displayed in your base currency, so a trip does not distort the monthly summary.

Can I share expenses with my partner?

There is no shared or family account: each account belongs to one person only. What there is is shared expense: you register what you paid, indicate the split, and the app keeps track of what they owe you. For the common fund of two people, each one keeps their own and splits what’s due.

In a nutshell

SpendTrack is AIsToHire's mini personal expense control app that reconstructs where your money goes based on the statement you export —without ever asking for access to your bank—, categorizes each expense into 14 categories, detects your subscriptions, and alerts you before you exceed your budget. It's at SpendTrack — Personal Expense Control Without Bank Connection Start with what yields the most: export the last three months, import them, and check the list of subscriptions. There's almost always something there that you've been paying for a year without using.

Export your bank statement from the last three months and upload it to SpendTrack: in one afternoon you'll know where your money goes, how much your fixed expenses weigh, and which subscriptions you're paying for without using. No need to connect your bank or write anything down by hand.

Export your bank statement from the last three months and upload it to SpendTrack: in one afternoon you'll know where your money goes, how much your fixed expenses weigh, and which subscriptions you're paying for without using. No need to connect your bank or write anything down by hand.

Keep reading

↑ Back to top