You import transactions by attaching a bank or credit card statement (CSV or PDF) to Dora and asking it to import them. Dora reads the statement, matches it to the right account, and hands you a Transaction import card to review. Nothing is imported until you approve that card.
This is for clients whose accounts aren't linked through account aggregation, or when you have history that predates the link. If the account is linked, transactions flow in automatically and you don't need to import anything.
How to Import
Open the client and start a chat with Dora.
Attach the statement — a CSV or PDF export from the bank or credit card. Tell Dora to import the transactions.
Confirm the account. Dora figures out which Kerdora account the statement belongs to using clues like the institution, account type, and last four digits. If it's unsure, it'll ask you to confirm — answer it. If the account doesn't exist in Kerdora yet, Dora will help you create it first.
Decide on categories and tags. If the statement has its own category or tag columns, Dora asks how to map them to your Kerdora categories and tags before importing. If it has none, skip this — Kerdora auto-categorizes after import.
Review the Transaction import card. Dora produces a card showing what's about to be imported.
Click Import on the card to bring the transactions in, or Reject to cancel.
Reading the Transaction Import Card
The card is the approval step. It shows:
A status badge — pending until you act, then approved or rejected.
The transaction count and the date range covered.
Inflows, Outflows, and Net totals so you can sanity-check the statement was read correctly.
The source file(s) the import came from.
Import and Reject buttons.
Check the totals against the statement before approving. If Inflows/Outflows look wrong (often a sign of how the file represents debits and credits), reject it and re-run rather than importing bad data.
What Happens After You Click Import
Approving kicks off a background job. The card updates in place with a visible progress phase and a progress bar so you can watch it run without leaving the chat:
Queued — Waiting for a worker to pick it up.
Classifying — Dora is matching each row to a category.
Importing — Transactions are being written into the client's ledger. The bar fills as rows complete.
Complete — Everything imported. Transactions are live on the Spending page.
Failed / Needs attention — Something blocked the import or some rows couldn't be classified. The card explains what to look at; you can usually retry or reject and re-run.
Imported transactions appear on the Spending → Transactions tab as they finish, auto-categorized with the AI source label. From there you clean them up like any other transactions: fix categories, hide transfers, and set up rules for the recurring ones.
Until you approve the card and the job finishes, nothing is in the client's file. Rejecting the card imports nothing.
Amounts: How Signs Work
Kerdora expects inflows positive, outflows negative. Dora handles this when it reads the file:
Money in (deposits, payroll, refunds) is positive.
Money out (purchases, payments) is negative.
Separate debit/credit columns get reconciled into one signed amount.
Accounting-style parentheses are treated as negative.
The Inflows/Outflows/Net totals on the card are your check that this came out right. If outflows show as inflows, the signs were misread — reject and try again.
Tips for a Clean Import
CSV beats PDF when the bank offers both. A clean CSV export imports more reliably than a scanned or formatted PDF.
Import the full statement period rather than a few rows, so the date range and totals are meaningful.
Watch the size. Very large statements (around 8,000+ rows in one go) can hit the agent's output-token limit while Dora is setting up accounts and categories. If you've got years of history, split it into a few smaller statement files and import them one at a time.
Don't double-import. If an account is already linked via account aggregation, you usually don't need to import — those transactions are already flowing in.
Map categories once. If your statement has good category data, mapping it during import saves cleanup later. If it doesn't, let Kerdora auto-categorize and use rules afterward.
Set up rules after the first import so the next statement for that account is mostly categorized on arrival.
Related Articles
The Spending Tab: Transactions, Cashflow, Categories, Tags, and Rules — the full picture.
Categorizing and Filtering Transactions — cleaning up after import.
Spending Rules: Automatically Categorize Transactions in Bulk — so the next import is hands-off.