Can OCR Read Business Cards Accurately?

OCR does two jobs on a card: it reads the characters, then it guesses which line is the name, which is the job title, and which number is the mobile. Reading the text is usually reliable — sorting it into the right fields is where most errors happen. This guide explains what OCR gets right, where it goes wrong, and how to get a result you can trust.

What OCR gets right

On a plain card — dark text on a light background, standard font, generous spacing — OCR is very good. The fields it handles best are the ones with a distinctive format:

• Email addresses and websites: their format is so distinctive that they are rarely misread, and a mistake is usually obvious at a glance.
• Phone numbers: digits in clear print come through reliably, especially when they carry a country code.
• Company names in plain type: a single line of ordinary text in a standard font is the easiest thing for OCR to read.
• Job titles on their own line: when the layout is simple, the title lands in the right field without trouble.

For background on how the technology works, see what OCR is and how it works. The short version: the software finds the text in the image, recognises the characters, and then uses the layout to decide which field each piece of text belongs to.

Where OCR goes wrong

The errors fall into predictable patterns. Knowing them tells you which cards to double-check:

• Several numbers or addresses: OCR may read them all correctly but put them in the wrong fields — the mobile in the office slot, the office in the fax slot, or a second address overwriting the first.
• Stylised or script fonts: decorative lettering can be misread, especially in names, where there is no dictionary to correct against.
• Busy or artistic designs: text over images, patterns, or gradients is harder to separate from the background, and characters get dropped or invented.
• Glossy or metallic finishes: glare from foil, gloss, or embossing can hide characters entirely — see how to remove glare from glossy photos.
• Very small type: the fine print at the bottom of a card — a second office address, a registration number — is the first thing to come back garbled.
• Low contrast: light grey text on white, or a coloured font on a coloured background, is hard for OCR to separate from the paper.
• Logos that look like text: a monogram or a lettermark can be read as a stray character and inserted into a field.
• Handwritten notes: anything written on the card after printing — a mobile number added in pen, a reminder — is usually ignored or misread.
• Bilingual cards: a card with two languages can confuse the language detection, and the wrong side may be read. Scan the side in the language you want stored.

How to get a more accurate read

Most scanning errors come from the photo, not the software. A few habits make a big difference:

1. Scan in good, even light and avoid overhead glare on glossy cards. Move the lamp to the side instead of shooting under a ceiling light.
2. Fill the frame. The card should take up nearly the whole image — small text is the first thing OCR misreads.
3. Hold the phone flat and parallel to the card. Skewed text is harder to recognise than straight text.
4. Scan one card at a time. A single card per image gives the software the best chance of identifying the fields correctly.
5. Set the right language if the card is not in English or carries accented characters — most OCR tools let you pick the language before scanning.
6. Scan the same day you receive the card, while you still remember the conversation. If a field looks wrong, your memory is the backup.

The 10-second check that catches nearly every error

Even good OCR benefits from a few seconds of human review. Checking the name, email, and mobile number before saving takes far less time than typing a card from scratch, and it stops small errors from reaching your follow-up emails. Read the three fields against the card, letter by letter, and pay special attention to the characters OCR confuses most: 0 and O, 1 and l, 5 and S, and rn read as m. If two numbers are listed, confirm each one is in the right field — a mobile saved as an office number is the most common field-swap error. The full save-as-contact workflow is in how to scan business cards and save them as contacts.

When to type the card in by hand

Sometimes manual entry is simply faster. Type the card yourself when it is heavily designed — a dark card with reversed-out text, a vertical layout, or a card that is mostly a photograph with text overlaid. The same applies to cards in a script you cannot read, cards with heavy embossing, and any card where the first scan comes back with more than two wrong fields. A minute of typing beats five minutes of correcting a bad read, and for a contact you will actually call, the certainty is worth it.

Frequently asked questions

Can OCR read business cards accurately?
Yes, on clean standard cards — independent tests put field accuracy above 90%. But 15–25% of fields can still carry an error on tricky layouts, so always check the name, phone number, and email before saving.

What does OCR get wrong most often on business cards?
Field assignment, not reading: it reads the text correctly but puts the mobile in the office slot or swaps two addresses. Stylised fonts, glossy glare, small type, and low contrast cause the most misreads.

How can I improve OCR accuracy on business cards?
Scan in good, even light with no glare, fill the frame with the card, hold the phone flat and parallel, scan one card at a time, and set the correct language. Most errors come from the photo, not the software.

Should I trust OCR output without checking it?
No. A 10-second check of the name, email, and phone number catches nearly every error. A wrong digit in a phone number is worse than no contact at all, because you will only discover it when the call fails.

Is manual entry ever better than OCR for business cards?
Yes — for heavily designed cards, dark cards with reversed-out text, vertical layouts, or any card where the first scan comes back with more than two wrong fields. A minute of typing beats five minutes of correcting a bad read.

Related guides

• How to Scan Business Cards and Save Them as Contacts
• What Is OCR and How Does It Work?
• How to Scan Documents with Your Phone Camera