Retrieve the SEK Amount, BankID Recipient, and Meddelande from a Swish Screenshot
Manually logging each Swish payment from a screenshot into Fortnox or Visma takes 2-3 minutes of retyping — this extracts the amount, recipient name, and notes in 5-10 seconds per image, regardless of which of Sweden's 20 partner banks the sender uses.
5-10s per screenshot · Up to 99% accuracy on printed text
What You Can Pick from a Swish Confirmation Screen
Name the fields you want — the AI locates each value by understanding its label context, not its pixel position.
Exclusively Swedish krona — no multi-currency ambiguity.
The legal name from BankID, tied to the personnummer — not a changeable handle.
07XX-XXX XX XX format — prefix and last four digits help match payments.
Carries invoice references with Swedish characters (å, ä, ö) — read holistically, not pixel-by-pixel.
A Legal Name, a Masked Number, and a 50-Char Message — What Makes Swish Different
Swish's data structure is unlike Venmo or PayPal — and that difference matters when you're automating screenshot extraction for bookkeeping.
The Recipient Is a Legal Name
Swish shows the recipient's BankID-registered legal name — not a @username or email. First-time users on forums regularly ask why it shows a full name. Traditional OCR dumps the name as raw text without context — you can't tell whether "Anna Andersson" is the sender or the recipient.
Swedish Characters in the Notes Field
The Meddelande field carries invoice references with å, ä, ö. Traditional OCR has a documented weakness on diacritics — JPEG compression blurs the dots on an ö, turning "för" into "for" without an error flag.
SEK and Masked Phone Confuse Dump OCR
SEK-only amounts have no other currency context — OCR can't tell if a number is the amount or a reference code. The masked phone (07XX-XXX XX XX) comes out as a jumbled string with dashes and X's, useless for matching.
Label-Aware Reading Picks the Right Name
Type Recipient Name as a column. The AI reads the label beside each name — it knows which is the recipient by its position relative to the "till" label and phone number block.
Holistic Reading for Swedish Characters
The model reads "för" as a complete word — not by decoding each dot on the ö separately. å, ä, ö are recognised as whole characters. Your Meddelande column gets clean Swedish text without silent OCR substitution.
Semantic Context Separates Amount from Noise
Define Amount (SEK) and the AI looks for the large centred figure paired with "SEK" or "kr" — not every number. The phone number returns in its masked format exactly as displayed.
From a Swish Confirmation to a Fortnox-Ready Row in Three Steps
If you receive 20-30 Swish payments a month for freelance work, here's how a batch of screenshots becomes a spreadsheet ready for Fortnox or Visma.
Upload the Screenshots
Drop a batch of Swish payment confirmations — JPG, PNG, or WebP, any Swedish bank, any app version. Swish's layout is consistent across all 20 partner banks. Forwarded images still work; the notes field may need a quicker spot-check.
Name the Columns
Define Amount (SEK), Recipient Name, Phone, Meddelande, Date. One column set for every screenshot in the batch — the AI identifies the BankID-linked name, reads Meddelande with Swedish character support, and separates the SEK amount from other numbers.
Export and Import
Export to Excel or CSV — five columns, one row per screenshot. Import into Fortnox, Visma eEkonomi, or Bokio via SIE format. Swedish accounting law (Bokföringslag) requires seven-year records; a structured spreadsheet replaces manual copy-paste from each confirmation screen.
When Swish Screenshot Extraction Works Best — and When It Doesn't
When It Works Best
Direct Swish app screenshots — high-contrast machine-rendered text at up to 99% accuracy for the amount and name.
Meddelande with Swedish characters — å, ä, ö are read holistically, accuracy comparable to English.
Batch with MobilePay — same field structure, one column set for both Nordic apps.
When to Be Cautious
Compressed/forwarded screenshots — WhatsApp or Messenger reduce image quality; Meddelande with Swedish characters may need spot-checking.
Outdated Swish app versions — Getswish AB has redesigned the layout multiple times; legacy screenshots may need a quick first-row check.
List/history views — extract from individual confirmation pages rather than multi-transaction screens for best results.
Frequently Asked Questions
Can I extract the recipient name even though it's a BankID-registered legal name?
Yes. The AI reads the Recipient Name by its position relative to the phone number and amount. Unlike a @username that changes, the BankID-linked name is tied to the personnummer — a stable identifier you can rely on for records.
Will the extraction handle Swedish characters like å, ä, ö in the Meddelande field?
Yes. The AI reads "för" as a complete word — not by decoding each dot on the ö separately. å, ä, ö are recognised as whole characters. Accuracy matches English on standard screenshots. The 50-character limit makes spot-checking quick.
Can I still use the masked phone number to match payments?
The number is extracted as displayed — 07XX-XXX XX XX format. The visible prefix and last four digits are enough to distinguish regular senders. For reliable matching, the Recipient Name field (BankID-linked, unchanging) is usually the better key.
Can I batch Swish screenshots with MobilePay in the same export?
Yes. Swish and MobilePay share the same field structure — amount, phone number, notes. Upload them together with one set of column names. Each screenshot becomes one row in the same spreadsheet, regardless of which Nordic app generated it.
Can I convert the SEK amount to EUR or USD during extraction?
Swish is SEK-only. The Amount (SEK) column captures the exact figure. Add a Computed Column like Amount (EUR) (Amount (SEK) × 0.087) and the AI calculates the conversion using your rate during extraction — no separate step needed.
Deep dive into Swish screenshot field locations and extraction details: How to spot the amount and recipient on Swish screenshots — including the BankID-linked recipient name that trips up first-time users
Swish and MobilePay share the same field structure for batch processing across Nordic payment apps: MobilePay screenshot extraction — field structure and batch workflow