Skip to content

Turn PDF bank statements into clean transaction data.

Upload a statement and see every transaction as a row in seconds. No account, no card.

No account needed to try it. No card. Your file is stored in the EU, and never used to train anything.

A statement goes in. Transactions come out.

Bank statement · PDF

Northbridge Commercial

Account
Kestrel Joinery Limited
Period
01/08/2026 – 31/08/2026
Opening
£12,480.15
  • 04 Aug · CARD 4417 TOOLSTATION−£186.40
  • 07 Aug · BACS REDGRAVE TIMBER−£3,340.08
  • 12 Aug · FPS HOLLOWAY FITOUT LTD+£9,750.00

Closing balance£18,703.67

04/08/2026CARD 4417 TOOLSTATION£186.40£12,293.75
07/08/2026BACS REDGRAVE TIMBER£3,340.08£8,953.67
12/08/2026FPS HOLLOWAY FITOUT LTD£9,750.00£18,703.67

Three of the transactions on this statement, with the running balance intact. Invented bank, invented figures.

How it works

  1. 1. Upload

    Add PDFs, scans, images or documents.

  2. 2. Structure

    Datanem works out the columns from the page itself and turns the useful information into clean rows.

  3. 3. Use it

    Download the data as a spreadsheet, or carry on working with it in Datanem.

Create an account to customise the columns and save the design for future batches.

That worked. Now make it yours.

A free account is where you choose what gets extracted, instead of taking what the first read gave you.

  • Choose the exact fields

    Rename, remove or add columns until the table is the one you wanted, then extract against it.

  • Save the design and reuse it

    Next month's pile goes through the same columns without designing anything again.

  • Process whole batches

    Up to 250 files in one go, and one row per document at the end of it.

  • Export the finished dataset

    Excel and CSV, plus SQL, JSON, JSONL and Python for anything that has to load it automatically.

Start free

20 documents to start. No card.

For bookkeeping

Bank statements into a transaction table

A bank statement is a table that has been printed and then locked. Copying it out of a PDF gives you columns welded into one, dates split across lines, and a running balance that no longer runs. Downloading CSV directly from the bank solves it, right up to the point where the account is closed, the period is older than the portal keeps, or the statement belongs to a client who sent a PDF and nothing else.

Datanem reads the batch, works out the columns your bank actually prints, and gives you one row per transaction with the balance intact.

Columns you can expect

There is no fixed template for this document type. Datanem reads your own documents and proposes columns like these; inside an account you can rename, remove or add any of them and save the design for future batches.

  • Account name
  • Account number
  • Sort code
  • Statement period
  • Opening balance
  • Closing balance
  • Transaction date
  • Description
  • Reference
  • Type
  • Money in
  • Money out
  • Balance

Every bank lays it out differently

Some print money in and money out as two columns, some print one signed amount, some put the type in a code column and some fold it into the description. There is no built-in statement design, because a design built around one bank is wrong for the next one.

In an account, Datanem samples a few of your statements, proposes the columns it can see, and shows them to you before extracting the batch. Confirm once and the design is saved, so every later statement from that bank needs no setup. A single statement tried without an account is read straight away instead.

Numbers you can check rather than trust

Every extracted value is compared against the text of the document itself, and anything that does not appear there verbatim is flagged for you to look at. Where a statement prints a total, the design can carry a check that compares it against the sum of the rows.

Values are read as printed. Datanem does not categorise transactions, reconcile them or convert currency, all of which need rules that belong to your bookkeeping rather than to a transcription tool.

Financial data, stored properly

A statement carries an account number and a spending history, which is about as sensitive as ordinary business data gets. Files and extracted data are stored in the EU, deleted automatically after your plan's retention window, and never used to train a model.

Retention runs from 30 days on Free to 365 days on Max, and you can delete a job yourself at any point before that.

Your structured data isn't locked inside Datanem

Once it's structured, the data is yours. Take it to a spreadsheet, a database or a script.

Excel— one sheet per table, ready to open.

  • CSV
  • SQL
  • JSON
  • JSONL
  • Python

Where your document goes

  • It is stored in the EU

    Uploads sit in an EU-only bucket and the database is EU-constrained. Neither leaves. Reading the document is done by a French provider whose standard endpoint does not commit to an inference location, so that step may run outside the EEA: the detail is on the security page.

  • It is deleted

    Uploads and the data drawn from them are removed automatically after 30 days on the free plan, longer on paid plans. You can delete a job yourself at any point before that, and the file leaves storage immediately when you do.

  • It never trains a model

    We have no training pipeline. Our extraction provider offers training as an opt-out and we have opted out, so your content is excluded.

  • Nothing here is a certification

    We are registered with the Information Commissioner's Office and the measures above are the ones actually in place. We hold no security certification and do not claim one.

Security & privacy · Privacy notice · Sub-processors · Data processing

What it costs

What can I do for free?
One document with no account at all, which is the trial at the top of this page. Create a free account and you get 20 documents to start with, then 10 a month after that, for ever. No card either way.
What uses it up?
Documents, not pages and not fields. A file of up to 10 pages counts as one document; a longer one counts as another document per 10 pages, so a 30-page statement counts as three.
What happens when the free documents run out?
Nothing is deleted and nothing is charged. Extraction stops until the monthly allowance resets, you buy credit, or you take a plan.
What is the first paid option?
Starter, at £9 a month for 150 documents. Monthly, no contract.
Is there pay as you go?
Yes. £0.20 per document with no subscription, bought from 25 upwards. Bought documents do not expire.
What are the limits on a file?
Up to 10 MB per file and up to 250 files in one batch. PDFs, Word documents, plain text and photographs; scans go through optical character recognition first.

Full pricing, including what each plan keeps and how long files are held.

Questions

Is there a built-in bank statement template?
No. Datanem reads a sample of your own statements and proposes columns from those, which you confirm or change before extraction. The design is then saved and reused for that bank.
How long can a statement be?
Up to 50 pages on Free, 75 on Starter, 120 on Pro and 150 on Max. One document covers up to 10 pages, so a 30 page statement counts as three.
Does it work on scanned or photographed statements?
Yes. Scans and images go through optical character recognition first. A text PDF reads more reliably than a photograph, so use the original file where you still have it.
Can it categorise the transactions?
No, and deliberately not. It transcribes what is printed. Categorisation depends on your chart of accounts, so it belongs in your accounting software or your own spreadsheet rules.

Turn PDF bank statements into clean transaction data.

No account, no card. One document, and you can see for yourself.

Choose a document

Other documents