Why scanned PDFs cost more to translate, and how to pay less

What a scan really is

A scanned PDF is a photo of a page saved inside a PDF file. To check, open it and try to select one word. If you can't, or the whole page highlights as a single block, there is no text in the file, only an image of text.

That matters because translators work on text. Their CAT tools (computer-assisted translation software such as Trados, memoQ, Phrase or OmegaT) split a document into sentences and keep the formatting around them. An image has no sentences to split. Before translation can start, someone has to turn the picture back into text and rebuild the page.

The American Translators Association (ATA) puts it plainly: editable formats such as Word or Excel are easier for a translator to handle, and scanned files take longer to translate, which may increase the cost.

The extra work hidden inside a scan quote

1. Text recognition

OCR (optical character recognition) software reads the letters in the image. Results depend heavily on the scan. Google's own tips for its free OCR ask for text at least 10 pixels high, pages the right way up, sharp images with even lighting and clear contrast, and common fonts. Faint photocopies, skewed pages and shadows all add errors.

2. Checking the recognized text

OCR mistakes are easy to miss: a 1 read as an l, a 0 read as an O, a comma dropped from a figure. Documents in Cyrillic or Greek can have look-alike letters swapped for Latin ones, which breaks search and spell-check. Someone has to compare the text against the scan line by line before translating, or the errors end up in the translation.

3. Rebuilding the layout

Even free OCR loses structure. Google says lists, tables, columns, footnotes and endnotes are not likely to be detected. A price table, a two-column contract or a form with boxes has to be rebuilt by hand so the translation has somewhere to go.

4. Layout work after translation

Once translated, the text usually changes length and has to be fitted back into the page. Translators call this desktop publishing, or DTP. ATA's guidance for reviewing translations says to agree ahead of time on charging for that work.

5. Hard-to-read content

Stamps, seals, signatures and handwritten notes are common on scanned records. ATA notes that handwritten documents may be hard to read, which slows the work.

Why the word count is harder to pin down

ATA explains that translation is often charged per word (or per character for languages written in characters), and that some translators charge per page or a flat fee instead. With a digital file, the word count is known before anyone starts. With a scan, there is no count until the text has been recognized, so the first figure is often an estimate.

Repeated text is another factor. CAT tools reuse earlier translations of the same sentence, and some providers price repeated or previously translated sentences differently. A sentence with one OCR error no longer matches its earlier version exactly, so a messy conversion can hide repetition you might otherwise benefit from.

Other things that raise a quote

ATA lists the main factors behind a translation price: document format, length, official certification, turnaround time, specialized vocabulary and the language pair. It adds that most professionals have a minimum charge, even for one or two pages, and that rush or weekend work usually costs more. A scan can add to several of these at once: a stamped birth certificate in a less common language, needed by Friday, combines most of them.

Prices vary widely between providers and language pairs, so compare written quotes rather than relying on a typical rate.

How to keep the cost down

1. Ask for the original file. If someone sent you a scan of a document they created, ask whether they have the Word file or a digital PDF. That one email can remove most of the extra work.

2. Scan well if you only have paper. Lay the page flat, keep it straight, use even light and avoid shadows. Include every page, and the back of any page with stamps or notes.

3. Send one clear file, in order. A jumble of phone photos takes longer to sort than a single PDF.

4. Mark what doesn't need translating. Repeated letterheads, page numbers or sections the receiving office doesn't need.

5. Do the conversion first, or pay someone to. You can convert the scan yourself (see how to turn a scanned PDF into an editable Word document) and send both the scan and the draft text. Ask whether the translator wants your version or prefers to do it.

Questions to ask before you accept a quote

• Is the price per word, per page or a flat fee?

• Is the word count from the scan an estimate, and what happens if the real count differs?

• Does the quote include text recognition and proofreading against the scan?

• Is layout work (DTP) included, and up to what level of matching the original?

• How will stamps, seals, signatures and handwriting be handled?

• If you supply a converted Word file, does the price change?

A sample comparison

Here is a synthetic example. Northgate Logistics needs a three-page supply agreement translated. Version A is the original Word file. Version B is a phone scan of the signed printout, with two columns, a price table and a company stamp. Same words, same deadline. For version B, the translator has to recognize the text, check it against the image, rebuild the table and columns, and fit the translation back into that layout. Expect the second quote to include more hours, or a higher page rate, for exactly that reason.

Turn the scan into a file translators can work with

EasyToDecode is building a converter for this step, launching soon. You'll upload the scan and get back an editable .docx with sentences kept whole, headings as Word styles and tables as real tables, ready for a CAT tool. It prepares the file; it doesn't translate it. See how it works or join the waitlist to hear when it opens.