How to track expenses without linking your bank account

If you have ever installed a budgeting app, you have met the screen: a list of bank logos, and a request to sign in to yours. Tap through it and the app fills itself in — every transaction, categorised, going back months. It is genuinely convenient, and it is the default because it works.

It is also not the only way, and for some people it is the wrong way. This post covers what that screen actually hands over, why the rules around it are less settled than the marketing suggests, and how to track spending without it — including the part most articles skip, which is that doing it manually costs you something real.

#What linking your bank actually hands over

When an app offers to connect to your bank, it usually is not talking to your bank itself. It uses an aggregator — Plaid is the best known — that sits between the two.

It is worth reading what that company says it collects, in its own words rather than the app's. Plaid's privacy policy names, among other categories:

• Login credentials — "username and password, account and routing number, or a security token"
• Financial data — "financial account name and number, balance, and transaction history"
• Transaction details — "amount, date, payee, type, quantity, price, location"
• Employment and income — "data about your income and employer", where payroll documents are connected
• Device data — IP address, location, hardware model, operating system and browser

Two of those tend to surprise people. The first is location on individual transactions — not just that you spent £40, but where. The second is that the list is broader than "the transactions I wanted my budget app to see": identity, device and, in some flows, employment.

On retention, the policy says data is kept "only as long as it is needed", with automatic deletion when a developer removes a connection — subject to exceptions for active connections, legal requirements, fraud prevention, and data that has been "aggregated, de-identified, or anonymized". I could not find a specific timeframe stated, and I am not going to invent one.

None of this is hidden. It is in a public policy anyone can read. The point is that very few people read it before tapping a bank logo, and the thing being agreed to is durable: a second company, with its own policy and its own retention rules, now holds a copy of your financial history.

#The rules around this are not settled yet

You might reasonably assume regulation has caught up with this by now. In the US, it has not — or rather, it did, and then it stopped.

The Consumer Financial Protection Bureau finalised a rule under Section 1033 of the Dodd-Frank Act, giving consumers rights over their financial data and setting terms for how third parties may access it. On the CFPB's own Personal Financial Data Rights page, the Bureau states that the compliance dates "were stayed by the court in Forcht Bank, N.A., et al. v. Consumer Financial Protection Bureau, et al." — and separately, that in August 2025 it issued an Advance Notice of Proposed Rulemaking "discussing possible amendments" to the rule, along with plans to extend the compliance dates.

So the rule exists, and it is not currently in force. What an aggregator may do with your data is, for now, mostly a question of what its own policy says — which is why reading that policy is not a paranoid exercise.

I am describing the US position because that is what the CFPB governs. Other jurisdictions differ, and I have not checked each one, so I am not going to generalise beyond what I can point at.

#The honest cost of doing it manually

Here is the part the "go manual, reclaim your privacy" articles tend to skip.

Automatic import does real work. It catches the £3.20 coffee you forgot, it reconciles itself, and it keeps going during the weeks you are not paying attention. Manual entry does none of that. If you do not type it, it is not there.

That has two consequences worth being honest about:

1. Your records are only as complete as your habit. A month where you stopped logging is a month with no data, not a month with approximate data.
2. You are the categoriser. That is an advantage when an algorithm would have filed your plumber under "Entertainment", and a burden on the evening you have nineteen receipts.

What you get in exchange is not only privacy. It is that you see each number as you enter it, which is a different relationship with your spending than reading a chart someone else assembled. Whether that trade is worth it depends entirely on why you are tracking in the first place. If you want a tax-ready ledger with zero effort, linking your bank is the honest answer. If you want to know where your money goes and would rather not hand a copy to a third party to find out, manual is viable.

#A method that survives past week two

Most manual tracking fails the same way: it is treated as bookkeeping, done in a weekly batch, and abandoned when a batch is missed. What works is making it small and immediate.

Log at the point of spending, not at the end of the day. The entry takes a few seconds while you are still holding the card. Recalling Tuesday on Thursday takes much longer and is where accuracy goes.

Use few categories. Eight to twelve is plenty. Thirty categories means a decision on every entry, and a decision on every entry is what makes people stop. You can always split later; you cannot recover entries you never made.

Reconcile weekly, not daily. Once a week, open your banking app, compare the balance to what your tracker says, and add whatever is missing. This is the step that makes manual tracking trustworthy — it converts "what I remembered" into "what actually happened", and it takes about five minutes. Note that you are reading your bank here, not connecting it.

Set up the recurring things once. Rent, subscriptions, utilities — anything with a fixed date and a predictable amount should post itself. These are the entries most worth automating, because they are the ones you are most likely to treat as background and forget.

Decide what to do about cash. Cash is where manual tracking quietly breaks, because there is no statement to reconcile against. The simplest workable rule is to log the withdrawal as spending in one category and stop tracking beyond that, rather than pretending you will itemise it.

#What to look for in a manual tracker

If you are choosing an app rather than a spreadsheet, these are the things worth checking before you commit a few months of data to it. None of them are specific to any one app.

• Does it have a server at all? "We don't sell your data" and "we never receive your data" are very different claims. An app with no backend cannot leak what it never had.
• Can you get your data out? CSV export, available on the free tier, not held behind an upgrade. This is the single best defence against a bad choice, because it makes leaving cheap.
• Is it a subscription? A budgeting tool that charges monthly is itself a recurring expense, which is at minimum ironic. One-time purchases exist.
• What happens to a free tier over time? Look for what is capped. Capped budgets is a limit; capped transactions means the app becomes unusable exactly when your history becomes valuable.
• Is the ledger encrypted on the device, and can you lock the app? If the data lives only on your phone, your phone's security is the whole of the security.
• Does it need permissions that do not match the feature? An offline tracker asking for broad message or contact access deserves a reason.

#A note on the app I build

I write these posts as the person who makes Baaki, so take the following as disclosed interest rather than a neutral recommendation.

Baaki is a manual expense tracker for iPhone and Android with no bank linking — not as a feature that is coming later, but because the app has no network client of its own to link with. The ledger sits in an encrypted database on the device. There is no account, so there is no server holding your records.

It does the things described above: recurring bills post themselves, CSV import and export are both in the free tier, and a "safe to spend" figure holds back reserved bills and money set aside for goals, so the number on the home screen is one you can actually use. Pro is a one-time purchase rather than a subscription.

It is not the only app that works this way — several others appear in any search for this topic, and some of them are good. The checklist above is written so you can judge them yourself, including against mine.

#Frequently asked questions