How to Automate Bank Statements with Dropbox

What the Dropbox Connection Does

Connect Dropbox to Bankflow and a folder becomes a pipeline. Statements that land in it are picked up automatically, parsed into transactions, and written back as a spreadsheet. Nobody uploads a file, and nobody downloads one.

Dropbox reads and writes, which is what separates it from a one-way export. It can be the place documents arrive and the place results are delivered, in the same folder if you want, and that symmetry is the reason it is usually the simplest connector to start with.

A Folder Is the Whole Interface

You browse to the folder rather than typing its path, so there is no mismatch between what you meant and what got saved. After that the folder is swept roughly every fifteen minutes, and there is a Run now button for when waiting is not appealing.

Each watched folder keeps its own sync cursor. One folder falling behind never causes another to re-import, and nothing is picked up twice. You can watch several folders in one workflow, because in practice statements rarely all arrive in one place.

What makes this shape work is that everyone already understands it. A client who will never log into a finance tool will happily drop a PDF into a shared Dropbox folder, and that is the entire onboarding.

Connecting Dropbox

• Connect the account from the connected accounts screen and approve the Dropbox consent screen.
• Create a workflow and add the folder to watch as its input.
• Say what the documents are: bank statements, which go into the ledger as transactions and balances, or invoices and receipts, which are read as line items and totals and deliberately kept out of the ledger.
• Add an output: write a file back to a folder, post to a signed webhook, or both.
• Turn it on, and use Run now once to confirm the path works end to end.

Dropbox permissions are granular, and Bankflow asks for exactly four: read who the account belongs to, list a folder, fetch a file, and write one back. That is the complete set the connector uses, and none of it extends beyond what the workflow does.

More Than One Dropbox Account

A personal Dropbox and the practice's Dropbox are two separate connections, told apart by the account each belongs to. Connect both, and each output names the account it writes to.

This is the arrangement most bookkeepers end up with: client documents arriving in one account, finished exports written into another that only staff can see. Reconnecting a connection whose token expired updates it in place rather than leaving a dead credential lying around.

What Comes Back, and Where

Results are written as CSV, XLSX, or JSON. The columns and the export presets are the same ones the manual export uses, so a file delivered into Dropbox and a file downloaded from the transactions page are identical.

An output defaults to the folder the file came from, which gives you PDFs and spreadsheets side by side. Each output can name its own destination instead, including a different account or a different cloud, so nothing forces the two halves to live together.

If you write results into the folder you are watching, the loop that suggests itself is already handled: Bankflow remembers every file it writes and skips its own output on the next sweep. Without that, a delivered CSV would be re-imported as a new document forever, and you would pay for the privilege.

Keeping the Connection Alive

Dropbox issues a long-lived connection only when it is explicitly asked to, and Bankflow asks. It is a small detail with a loud failure mode: get it wrong and the connection works for about four hours and then stops, permanently.

Tokens are encrypted at rest and refreshed in the background. If a connection genuinely cannot be renewed, it says so on the connections screen rather than failing quietly on the next sweep and leaving you to notice weeks later that a folder stopped filling.

When a Delivery Does Not Land

The workflow history reports parsing and delivery separately, because they fail independently. A statement can parse perfectly and still not reach the folder, and a single combined status would hide that.

A failed delivery never fails the import. By the time delivery runs, the statement is parsed, billed, and in your ledger. The failure is recorded rather than swallowed, and it can be retried.

The same logic governs Remove from Bankflow once sent, the option for pipelines where the copy here is a duplicate of something now living in your accounting system. It is honoured only when every output landed. One failed delivery and nothing is removed.

Dropbox, OneDrive, or Google Drive?

All three connect the same way and use the same workflow machinery. The differences that matter fit in a table.

The one asymmetry worth explaining is Google Drive. It cannot watch a folder because Bankflow connects to it under a non-sensitive scope that only reaches files the app itself created. The scopes that could watch a folder are classified restricted by Google and carry an annual third-party security assessment. Keeping the narrow scope was the better trade, so Drive receives results and does not supply documents. The Google Drive guide covers what that means in practice, and the OneDrive guide covers the Microsoft side.

You are not choosing one. A workflow can watch a Dropbox folder and write finished Google Sheets into Drive, because the connectors are independent and a workflow can span two of them.

Starting Small

The honest way to test this is with one folder and one month. Connect Dropbox, watch a folder holding a few statements, add an output back to the same folder, and press Run now. You will see within a minute or two whether the rows come out the way you want, and the history will tell you separately whether they parsed and whether they were delivered.

Fifty free page credits, no card required. If you want the background on the parsing itself first, start with how AI bank statement analysis works, and if the transactions ultimately need to reach an accounting package, the integrations hub covers which file each one accepts.