Tracking Your Money Without Linking Your Bank Account
You don’t need to link a bank account to know what you spend and what you’re worth. Your bank already gives you everything a tracker needs: a statement, every month, with every line on it. The question is only how much of the work you do by hand.
There are three ways to do it: a spreadsheet, a manual-entry app, or an app that imports your statements. I’ve used all three. In this post I compare them honestly, for spending and for net worth, and then explain why, after six years of trying, I won’t give any app my bank password again.
The ways to track without linking
Whichever you pick, you’re tracking two different things, and they ask for different work:
- Spending needs every line: what left your accounts, each one categorised.
- Net worth needs one figure per account at each month end: what you have, less what you owe.
A method can be good at one and poor at the other.
Spreadsheet
Once a month you download each account’s statement, paste or type the lines into a sheet, give each one a category, and note every balance on the last day of the month. That was my routine for years, before I built an app around it.
- Effort: an hour or two a month once the sheet is set up, more in the first months while you build the categories and formulas.
- Spending: as detailed as you want it. Every line is there, and you choose how to group them.
- Net worth: easy. One row per account, one column per month, a sum at the bottom. Loans and a mortgage are just negative rows.
- What it gets right: it’s free, it’s yours, it works with any bank, and nothing leaves your computer unless you put it in the cloud yourself. You can chart anything.
- Where it breaks: everything is by hand, so a pasted range that misses the last row or a formula that stops one cell short gives you a wrong total that looks right. Categorising the same supermarket every month gets old. And a spreadsheet doesn’t remind you: skip a month and the gap stays.
Manual-entry app
A budgeting or net-worth app where you type each purchase as you make it, or at the end of the day, and update account balances yourself.
- Effort: a minute or two a day, every day, plus a balance check now and then.
- Spending: complete, if you never forget. Cash is no different from a card, and logging a purchase as you make it is a moment to ask whether you want it.
- Net worth: usually a list of accounts whose balances you type. Fine for a few accounts, tedious for investments that move every day.
- What it gets right: no bank login, works with any bank and any currency, and the daily habit makes spending very visible.
- Where it breaks: the record is only as good as your memory. Miss a few coffees, a direct debit or a refund and the totals drift from what your bank says, and finding the difference later is painful. Check where the app keeps your data, too: some keep it on your device, others on their own servers. I compare the apps for Mac, linked and not, in the best personal finance apps for Mac.
Statement import
An app that reads the statement file your bank gives you (usually CSV, Excel or OFX) and turns it into lines, categories and balances. You still download the statements, but you no longer type anything from them.
Lume, which I make, is one example. Once a month I download each account’s statement and import it. The first time, I tell Lume which column is the date and which is the amount, and it remembers that for the bank. Each line gets a category, from rules made in earlier months, so I only choose for merchants it hasn’t seen. Transfers between my own accounts pair up and count as neither income nor spending. Every account’s month-end balance comes from its statement, and what no statement gives (cash, the value of a car) I type. The file is read on my Mac or iPhone and never leaves it.
- Effort: under an hour for the first month, mostly mapping columns and teaching the app your merchants. After that, around ten to thirty minutes a month, most of it downloading the files.
- Spending: every line the bank has, categorised mostly by rules. Cash still has to be typed.
- Net worth: balances come from the statements, so they match the bank. Investments need your broker’s trades or holdings, and assets with no statement (a house, a car) are typed.
- What it gets right: the numbers are the bank’s, not your memory’s, with no bank login and no connection to keep alive. The effort is monthly, not daily.
- Where it breaks: it is only as current as your last download, so it shows last month, not this morning. It depends on your bank offering a file an app can read: a scanned PDF won’t do. And downloading statements from five banks is still five logins a month.
The lines you can’t place
Whichever method you choose, you will hit this: a line that says CRD 4471 MKTPL*EU 1192 42.00 and you have no idea what it was. Statements describe the payment, not the purchase. The merchant’s legal name, a payment processor, a parent company: none of it matches what you remember buying.
It’s tempting to drop those into Other and move on. Don’t, at least not every time. Working out what a line was (checking your email for the receipt, your calendar for the day, the shop’s legal name) is the useful part. It’s how you find the subscription that renewed without telling you, the parking app that charges more than you thought, or the €42 that turns out to be a third order from the same shop this month. You only see where the money goes by following it.
That’s why I think tracking once a month beats tracking live: you sit down with a whole month of lines and look at each one you don’t recognise. I wrote up the routine in the monthly close.
Side by side
| Sheet | Manual | Import | |
|---|---|---|---|
| Effort | 1–2 hours a month | Minutes a day | 10–30 min a month |
| Spending | Every line, by hand | What you log | Every line, by rules |
| Cash | Typed | Typed | Typed |
| Net worth | Typed | Typed | From the bank |
| Up to date | Last month | Today | Last month |
| Breaks on | Formulas | Memory | No file |
| Bank login | No | No | No |
If you like building your own tools and enjoy the maths, a spreadsheet is hard to beat. If the daily habit is what keeps your spending in check, a manual-entry app does that better than anything. If you want the bank’s numbers without giving anyone your bank login, import statements.
Why I won’t give an app my bank password
Before I tracked from statements, I spent years with apps that linked to my bank. This half is opinion: mine, from that experience, and from reading what the companies in the middle say about themselves.
The connections never stayed connected
Linking an account is sold as the effortless option. In my experience it was the opposite.
- Links broke when a bank changed something. A bank would update its app or its interface, and the connection would stop working.
- The fix took weeks, sometimes months. I’d report it, or see it already reported by other users, and wait while the app and its data provider caught up with the bank.
- The gap was lost. Lines from those weeks often never came in at all, or came in partly. Sometimes nothing told me: the account just stopped syncing, and I found the hole in my history a month later when the totals didn’t match.
- I reconnected again and again. Even when nothing was broken, the link would expire and ask me to connect once more.
- Every reconnection meant the bank’s security steps again: the login, the code from the bank’s app or a text message, sometimes on each account at each bank.
I wasn’t the only one. In September 2023, Wallet by BudgetBakers lost its connection to Portuguese banks, and its App Store reviews in Portugal filled up with paying users waiting for it to come back. In October one wrote that the date the app gave for the fix had moved for the third time since September. In November another asked: “It’s been two months, is there any plans to fix this?” The reviews saying it worked again only arrived in early 2024.
Some of that is by design. In Europe, a bank must ask you for strong customer authentication (two of a password, your phone, your fingerprint) the first time an account information service reads your account. Under the original rules the bank could skip it for 90 days at a time, after which you had to authenticate again. A 2022 amendment (Article 10a, in force since 25 July 2023) raised that to 180 days; the European Banking Authority proposed the change to reduce the friction it caused. With five or six accounts at different banks, even every 180 days means a reconnection somewhere most months.
A monthly statement download asks for the same login, but it never silently stops, and when it’s done I have the whole month.
What the company in the middle holds and keeps
Many apps don’t connect to your bank themselves. They use an aggregator: a company whose business is connecting thousands of apps to thousands of banks. When you link an account, you’re trusting the app and the aggregator, and the aggregator is the one you never chose.
Plaid is the best-known one, and its end user privacy policy (effective 8 December 2025, read on 12 October 2026) is unusually clear about what it collects. It includes:
- Login data “when required by the provider of your account, like your username and password, account and routing number, or a security token”, and security questions and one-time passwords.
- Account data and every line: balances, account numbers and IBANs, and for each line “amount, date, payee, type, quantity, price, location”. Depending on how the bank makes data available, this “may include data from all accounts (e.g., checking, savings, credit card, and joint accounts) accessible through a single set of account credentials”, not only the one you picked.
- What it derives: “we may infer your geolocation, your annual income, or the type of account”.
And what it uses it for, besides running the service: to “develop insights” from what it holds, which “includes your transaction data”, and to develop new products, including to “train, test, and deploy artificial intelligence (‘AI’) systems”. Aggregated or de-identified data can be used “for any purpose permitted under applicable law”.
On keeping it: if the app removes its connection to your data, Plaid’s “systems are designed to automatically delete your personal data, subject to certain exceptions”. The exceptions include another app still connected through Plaid, fraud prevention, legal requirements, and data that has been de-identified. You can see and delete your connections through its Plaid Portal.
That is a privacy policy doing its job. But it’s also a description of a company holding your login or a token for it, years of your lines, and what it infers from them, across every app you have ever linked.
Plaid has also been sued over this. In Cottle v. Plaid (N.D. Cal., No. 20-cv-03056), the plaintiffs alleged that Plaid used “consumers’ banking login credentials to harvest and sell detailed financial data without their consent”, and that its login screens had “the look and feel of login screens used by individual financial institutions”. Plaid agreed to pay $58 million and to change its practices: deleting some data, minimising what it stores, and expanding its privacy policy. The court gave final approval on 20 July 2022. It was a settlement, so no court decided whether the allegations were true.
I have no reason to think Plaid is careless with security, and its current policy is more open than most. For many people, linking through it is a reasonable trade. It isn’t for me, because the problem isn’t whether it’s secure today. It’s that a company I never chose keeps a detailed history of my money, uses it for purposes of its own, and I have no way to check any of that from the outside.
Screen-scraping, open banking, and what “read-only” means
There are two ways an aggregator reads your account.
Screen-scraping: you give it your bank password, and it logs in as you. The US Consumer Financial Protection Bureau calls it “a still common but risky practice” (CFPB, 22 October 2024).
API access, what Europe calls open banking: you authenticate with the bank directly, and the bank gives the provider its own access. Under Article 67 of PSD2, the provider needs your explicit consent and may use the data only for the service you asked for. The US has no equivalent in force: the CFPB’s 2024 rule is being reconsidered, and a court stayed its compliance dates.
“Read-only” means something real, but narrower than it sounds. Through an API, the provider can read your balance and lines but can’t move money. That’s why open banking beats handing over a password. But:
- It limits what the provider can do to your account, not what it does with the copy. A full history of your spending is the sensitive part, and reading it is exactly what read-only allows.
- With screen-scraping, it’s a promise, not a limit. Whoever holds your password can log in as you. Plaid’s policy says your login data gives it “permission and authority to act on your behalf”.
- It doesn’t cover the app on top. Plaid’s policy “does not cover what developers of the apps you use do with your data”.
How I do it now
I download my statements on the first weekend of each month and import them into Lume. It reads them on my Mac or iPhone, so they never leave the device. Lume keeps my data encrypted there, and when I sync between the two, it goes through end-to-end encryption with a passphrase that never leaves my devices either, so the server holds a copy it can’t read. Lume does use the network for some things: valuing investments needs prices, and those lookups go through a server that sees the ticker and my IP address. Its privacy policy lists all of it.
None of that requires an app. A spreadsheet and a monthly habit got me most of the way for years, and they will for you. What matters is the habit: a month of real lines, looked at once. I wrote about why the bank connections pushed me to statements in the first place in how I built Lume, and Getting Started goes from an empty app to a confirmed month if you want to try it that way.
General information, not financial advice. Your taxes, pension and circumstances will differ from the examples here. More on that.