The Financial Information Gap: Turning a Bank Statement Into an Overview

Last March I opened my DKB export and counted: 312 lines. Dates on the left, amounts on the right, and in the middle text like PAYPAL *SPOTIFY AB 35314369001 or KARTENZAHLUNG 04.03. REWE SAGT DANKE 123456.

I could see that money had come in and that more money had gone out. I could not see what those 312 lines added up to per shop, since when the PayPal debit had been running, or whether REWE cost me closer to 200 € or 400 € a month. The bank has every one of those numbers, but it prints them as a list and the reading is up to you.

Why an unsorted bank statement is annoying

An unsorted statement costs you in small ways that add up. You keep paying for a service you stopped using two years ago, because its debit line looks like all the others. You have a feeling for what groceries cost you, but no number. When the rent goes up, you cannot say by how much your fixed costs rose, because you never had the old total. And in April you scroll through hundreds of lines to find the six that belong in the tax return.

None of that is difficult, but sorting 300 rows by hand is dull enough that nobody does it. That was my situation in Bernau, north of Berlin, with statements from two banks piling up. I wrote Pinke to do the sorting for me.

What Pinke does with the list

Pinke reads your bank export and puts a name of your choosing on every line. KARTENZAHLUNG REWE SAGT DANKE becomes Groceries, and once enough lines have names, Pinke draws a chart from them.

It currently reads exports from Revolut, DKB, Sparkasse, Postbank, Vivid and Trade Republic (CSV or PDF, depending on the bank). If your bank is missing, click Bank not supported? under the upload box on /ingest. A window called Convert any bank export with your AI opens with a ready-made prompt. You paste your export into ChatGPT or Claude, get a clean CSV back and upload that. Your statement goes to the AI service you picked, so check the CSV before you upload it; Pinke itself sends nothing to any AI.

Walkthrough: teach Pinke about your bakery

Say you buy bread every week at the same place. On the statement it looks like this:

04.03.2026   KARTENZAHLUNG BAECKEREI MOHN GMBH//BERLIN   -4,80 EUR
11.03.2026   KARTENZAHLUNG BAECKEREI MOHN GMBH//BERLIN   -5,20 EUR
18.03.2026   KARTENZAHLUNG BAECKEREI MOHN GMBH//BERLIN   -4,80 EUR

14,80 € in three lines that nobody will ever look at. Here is how you sort them once and then never again.

1. Upload the file. Go to /ingest. There is a dashed box that says "Click to upload or drag and drop". Drop the export in.

2. Look at what came back. A card shows four numbers: Bank, Rows, Days, Balance. Compare them with the file. If the card says 312 rows and your statement has 312 lines, the parser has read the whole thing.

3. Sort it. Click Analyze. Pinke runs its rules over every row and fills in group, category and subcategory wherever a rule matches.

4. Fix the bakery. Open one of the bakery rows in the table. A window opens with the transaction at the top. Click Create rule and you get a sentence with blanks:

> Set group ______, category ______, subcategory ______, and tags ______

Fill it in like a sentence: group Living, category Groceries, subcategory Bakery.

5. Make it count for all of them. Below the sentence sits a section called Rule actions with two checkboxes, both unticked when the window opens:

• Apply to similar transactions fixes the bakery rows already in the file. Pick a Match field (Payee, Merchant or Description) and Pinke lists the rows it will change before you commit.
• Save as future rule stores the mapping so that next month's bread sorts itself.

Tick both and click Save.

6. Keep the file. Click Save on the file card to store the processed data.

One correction covers every future visit to that bakery, provided you tick the boxes. They start unticked because I would rather have Pinke ask than quietly rewrite 40 rows while you are looking at a different one.

Where to look afterwards

Once the rows have names, the other pages become useful.

• /visualize/sunburst draws your spending as rings. The inner ring holds the big groups, the outer rings break them down. Click a ring to zoom into it and use the Groups / Categories / Subcategories buttons to set the depth. The page is titled Spenderella's Wheel, and clicking that title spins it. I could not resist.
• /visualize is the hub for the rest. The cards are called Flow Mo' Show, The Balancing Act, Pie Hard: Expense Edition, The Paper Trail, The Subscription Trap™, Rewards, Portfolio Overview and Tax Overview. The subscription one is where I found a 7,99 € debit for something I had stopped using a year earlier; The Silent Drain shows how to read that page.
• /rules lists every rule you made next to the built-in system rules. Add one with New rule. Each rule has a small magnifier button, Test rule against history, that tells you how many past transactions it would catch. Rules That Stick covers what makes a rule hold up over time.
• /merchants holds merchants and their aliases, for when the same shop appears under three spellings.
• /report lets you filter your transactions and click Create to turn the selection into a document.

About your data

Many budgeting apps want you to connect your live bank account. That gives a company standing access to everything you do with money, and quite often the pie chart it shows you is there to sell you a credit card.

Pinke has no bank connection; you export a file yourself and upload it, so you decide what Pinke sees and when.

"Private" gets used loosely, so here is the precise version: Pinke runs on a server. When you upload a statement, the file goes to that server, is processed there and is stored with your account. Your browser only uploads it.

What Pinke promises is narrower than "private", and you can check each point:

• No analytics, no tracking cookies, no ads.
• Your uploads are used for showing you your own analysis, and for nothing else.
• Your data is not sold.
• You can delete it. On /me there is Delete all files (keeps your profile) and Clear all data.

The last point matters most to me: if you can delete everything in two clicks, you do not have to take my word for the rest.

The point

You should not need to be a spreadsheet person to know where your money goes. You already have every number; what is missing is a readable version of them.

Export one statement and upload it at /ingest. Spend ten minutes naming your regular merchants. Then open the wheel and find out what you spent on bread this year.