This is the hardest conversion there is, and anyone telling you otherwise has not done many. A PDF does not contain paragraphs. It contains glyphs with coordinates, drawn in whatever order the layout software felt like drawing them. Turning that back into a flowing eBook is not extraction. It is reconstruction, done by a person, with the original open beside it.
A PDF is a description of a printed page. It records that a particular glyph sits at a particular coordinate in a particular font at a particular size. It does not record that those glyphs form a word, that the words form a sentence, or that the sentence continues at the top of the next column. All of that structure lives in your head when you read it, and nowhere in the file.
An EPUB is the opposite kind of object. It has no coordinates and no pages. Text flows, and the reader decides how wide the column is, how large the type is and where the lines break. A conversion between the two therefore has to invent everything the PDF discarded when it was made, which is most of what a document is.
That is why we say rebuilt rather than converted. The text is recovered, the reading order is reconstructed, the paragraphs are reassembled from lines, the print furniture is deleted, and the structure is applied by hand. What comes out is a genuine reflowable eBook. What comes out of a one-click PDF converter is your book put through a shredder and taped back together in the dark.
Text extraction pulls the glyphs out in drawing order. On a single-column page that is often close enough to reading order to look like success. On a two-column page it interleaves the columns, so line one of the left column is followed by line one of the right, and the result is a fluent-looking sentence that means nothing.
Then there is the hyphenation. Every line that ended in a hyphen in the printed book carries that hyphen as a real character. After the text reflows to a new width, the hyphens stay exactly where they were, appearing in the middle of sentences, in words that no longer break there. Nothing flags this. It simply ships.
A PDF has lines, not paragraphs. Deciding where one paragraph ends and the next begins means reading the indents, the spacing and the sense of the text, and no extractor does the last of those.
The running head, the folio and the footer were drawn on the page like everything else. Extract the page and they arrive in the middle of your prose, once every few hundred words, forever.
A footnote sat at the bottom of page 84 because the marker was on page 84. Remove the page and it is simply an orphaned block of small text, with nothing connecting it to anything.
Every stage below is part of the price. None of them can be skipped, which is why this conversion costs more than any other.
The text is recovered and then put back into the order a human reads it, which is frequently not the order the software drew it. Columns are unstitched, sidebars are separated from the body, captions are attached to their figures, and anything that crosses a page boundary is rejoined into a single continuous paragraph.
A PDF holds a stack of lines. We reassemble them into paragraphs by reading the indents, the spacing and the sense, then strip every line-end hyphen the printed layout baked into the words. This is slow, it is done by eye, and it is the difference between a readable book and a mess.
Running heads, folios, footers, rules, watermarks and the publisher's name at the foot of every recto. All of it was drawn onto the page as content, all of it will otherwise appear scattered through your prose, and all of it has to be identified and deleted.
Where the pages are images, optical recognition runs first and then a person proofs the result against the scan. Recognition is confidently wrong in specific ways: it swaps similar letters, mangles accents and invents plausible words. Only a reader comparing the two will ever catch that.
In a PDF a table is a set of drawn lines with words floating between them. There is no table. We rebuild it as a real table with header cells, or restructure it where it is too wide to survive a phone, and we re-extract every figure at the best resolution the file holds.
Once the content is real again, the eBook is built properly: semantic markup, a stylesheet written for the book, navigation generated from actual headings, then EPUBCheck to zero errors and a pass on physical hardware before you ever see it.
Six recent rebuilds, from clean digital exports to scans of books nobody had the source files for.
These are two different jobs wearing the same file extension, and the difference is the largest number in the quote. Open your file and try to select a sentence with the cursor.
The file was exported from a layout program, so real text is inside it. The characters can be recovered, which removes the recognition stage entirely. What remains is the reconstruction: reading order, paragraphs, hyphens, furniture, notes, tables.
Still a rebuild rather than an extraction, but a considerably cheaper one, and the text will be exactly correct because it was never guessed at.
There is no text in the file at all, only photographs of text. Optical recognition has to produce the characters, and recognition makes errors with total confidence. Every page then has to be proofed against the scan by a person before any structure is applied.
This is the most expensive conversion we offer, and it is honestly priced as a rebuild. It is also the only way a book with no surviving source files comes back to life.
Six failures that appear in every automated PDF conversion we are sent. None of them raises an error, and all of them reach the reader.
Eight elements of a printed page, what a rebuild does with each, and what happens when nobody does.
| Element | How we build it | Why it matters |
|---|---|---|
| The text layer | Recovered where it exists, recognised where it does not, then proofed and reassembled into paragraphs rather than left as a stack of lines. | Extracted text is a stream of lines with no paragraph boundaries. Reflowed as it stands, the book becomes one enormous undifferentiated block. |
| Reading order | Reconstructed by a person against the printed page, with columns unstitched, sidebars separated and cross-page paragraphs rejoined. | Software reads a PDF in drawing order. Drawing order and reading order agree only by luck, and a multi-column page is where the luck runs out. |
| Line-end hyphens | Every one is identified and removed, with genuine compound hyphens preserved by checking the word against the surrounding text. | Left in place they become permanent misspellings. This is the single clearest signature of a PDF conversion that nobody checked. |
| Running heads and folios | Detected by position and repetition, then deleted from the content entirely before any structure is applied. | They were drawn on the page like any other text. Extraction cannot distinguish them from your prose, so they appear in it, endlessly. |
| Footnotes | Unanchored from their physical page, matched to the marker they belong to, and rebuilt as bidirectional links with popup semantics where supported. | A note is only a note because of where it sat. Remove the page and the relationship is gone, leaving small orphaned text with no purpose. |
| Tables | Rebuilt as real tables with header cells, from the words and the drawn lines, or restructured into a stacked form where they cannot fit a narrow screen. | A PDF table is not a table. It is rules and floating words. Extraction produces a jumble of cell contents in whatever order they were drawn. |
| Images and figures | Extracted at the highest resolution the file holds, re-exported for high-density screens, reattached to their captions and given alt text. | Print PDFs hold images in a print colour space at print resolution. Passed through untouched they are heavy, wrongly coloured and often separated from their caption. |
| Scanned pages | Recognition runs, then a human reads the output against the scanned image before any of it is trusted, page by page. | Recognition does not fail loudly. It substitutes characters and produces a fluent, wrong book. Nothing but a person reading will find it. |
The printed page is the only authority. Six checks decide whether the eBook still agrees with it.
Someone reads the rebuilt text beside the PDF, page by page. Not skimmed and not sampled. On a scanned title this is the largest single cost in the job, and it is the only thing standing between you and a book full of confidently invented words.
Chapters, headings, notes, tables, figures and captions are counted in the PDF and counted again in the EPUB. If the two numbers differ we know exactly how much has gone missing before anyone starts looking for it.
The whole text is searched for surviving line-end hyphens, broken ligatures, mangled accents, reversed quotation marks and the specific substitutions that recognition software likes to make. These hide easily and this pass finds them.
The rebuilt book is opened on a phone at a large type size. Anything still carrying the geometry of the printed page reveals itself immediately: a stray folio, an orphaned note, a table refusing to fold.
The derived Kindle file is checked, and the EPUB is opened on physical e-ink. Legacy titles rebuilt from print tend to be long, and long files are where older hardware starts complaining about memory and page turns.
EPUBCheck to zero errors, then each retailer's own ingestion check. A rebuilt book has to be indistinguishable from one that was born digital, and this is where that claim is tested.
A book that has been genuinely rebuilt, and the evidence that the rebuild is faithful.
A validated, hand-coded reflowable eBook, indistinguishable from one built from a live manuscript. Real headings, real paragraphs, working navigation, a stylesheet a human wrote. Ready for every store that takes EPUB, which is all of them.
Built from the same rebuilt source and confirmed on Amazon's rendering path before delivery. Titles recovered from print tend to be long and note-heavy, and Amazon's ingestion has opinions about both, so we look first.
The recovered book as an editable manuscript with real heading styles. If your source files are lost, this is the closest thing you will ever have to getting them back, and it is what a second edition or a new print run starts from.
What was in the printed book, what came across, what was changed and why. On a scanned title it also records the recognition pass and the proofing. This is how you know the rebuild is honest rather than merely finished.
Every figure, photograph and diagram pulled from the PDF at the best resolution it holds, converted out of print colour, re-exported for screens and named consistently. Useful again the moment you want a cover or a sample.
XHTML, stylesheet and package document, unzipped and commented. A rebuilt book should not be trapped in another opaque file, which is exactly the situation you came to us to escape.
Five stages. The middle three are where the money goes, and they cannot be hurried without lying to you.
We open the file and find out what it really is: digital or scanned, one column or two, how much furniture, how many notes, how many tables. Try selecting text yourself before you send it, because that one test predicts most of the price.
Characters are extracted where they exist and recognised where they do not. On a scan, this is followed immediately by a human proofing pass against the page images, because unproofed recognition is not text, it is a guess.
Reading order rebuilt, columns unstitched, paragraphs reassembled from lines, hyphens removed, running heads and folios deleted, notes reattached to their markers, tables rebuilt. This is the job. Everything else is comparatively quick.
Semantic markup applied, stylesheet written, navigation generated from the real headings, images re-exported and placed. Only now is there a document to format, which is why nobody should quote this as a formatting job.
Counts reconciled against the printed book, text read beside the original, EPUBCheck to zero errors, hardware pass, then delivery with the fidelity report and the recovered manuscript.
Pricing follows the word count, because every word has to be recovered, reassembled and read back against the page. Whether the PDF is digital or scanned then moves the number more than anything else.
Required steps are marked with *. Optional extras can be removed from the summary with the cross icon.
What kind of book the PDF holds. This decides how much of the printed structure has to be rebuilt rather than simply recovered.
The approximate word count. If you only have a page count, multiply by 300 for a rough figure and we will confirm it from the file itself.
Digital or scanned, single column or multiple. This is the biggest number in a PDF conversion quote and there is no way to disguise it.
How much of the page is not plain prose. Notes, tables and references were all anchored to paper, and paper is what we are taking away.
How many figures and photographs are inside. Print PDFs hold these at print resolution in a print colour space, and both have to be corrected.
Standard is the pace at which a book can be rebuilt and proofed. The faster tiers buy priority and longer days, never fewer checks.
Two rounds are included. On a rebuilt book these are usually about presentation, because the fidelity questions were settled during verification.
We rebuild structure, never wording. The text we recover is the text that was printed, typos and all, and we will not silently correct your book while converting it. Editing, proofreading, cover design and uploading to retailer accounts are separate services.
One of these takes four seconds. The other takes days, and it is the only one that produces a book.
| What you are comparing | Automated PDF Extraction | Book Formatting Experts |
|---|---|---|
| Reading order | ✕ Whatever order the page was drawn in. Two columns interleave line by line and produce sentences nobody wrote. | ✓ Reconstructed by a person against the printed page, with columns unstitched and cross-page paragraphs rejoined. |
| Hyphenation | ✕ Line-end hyphens survive as real characters and reappear in the middle of words after reflow, permanently. | ✓ Every one removed, with genuine compound hyphens preserved by checking each word against its context. |
| Running heads and folios | ✕ Extracted as content. Your book title and a page number now interrupt the prose a few hundred times. | ✓ Detected by repetition and position, and deleted before any structure is applied. |
| Scanned pages | ✕ Recognised and shipped. The output is fluent, plausible and wrong, and nothing tells you which words were invented. | ✓ Recognised, then proofed page by page against the scan by a person, because that is the only check that works. |
| Footnotes | ✕ Left as orphaned blocks of small text, unlinked, wherever the page they lived on used to be. | ✓ Matched to their markers and rebuilt as bidirectional links, with popup notes where the reading system supports them. |
| What you end up with | ✕ A file that opens, reads badly, and cannot be fixed without doing the whole job properly anyway. | ✓ A genuine reflowable EPUB, plus an editable manuscript, plus a written record of everything that changed. |
Six recurring cases, each one with a characteristic way of resisting conversion.
Rights have reverted, the publisher has sent a print PDF and nothing else, and the manuscript files went with a laptop three moves ago. The whole book has to be recovered from the page. It is the most satisfying work we do, because a title genuinely comes back into print.
Two hundred footnotes tied to the pages they sat on, tables built from drawn rules, a bibliography and an index of dead numbers. Nothing in it converts. Every relationship in the book has to be rediscovered and then rebuilt as a link that works in both directions.
Multi-column pages, pull quotes, sidebars, charts and infographics, all positioned rather than structured. Extraction weaves the columns together and the document becomes unreadable in the first paragraph.
Ingredient columns beside method steps, both floating over a full-bleed photograph. Recovering it means working out which words belonged to which recipe, and then deciding whether the book should reflow at all.
Often a scan rather than a digital file, with photographs, letters and captions. Recognition struggles with old type, so the proofing pass is long, and the photographs need real work before a screen will do them justice.
The hardest thing on this page. Line breaks carried meaning and the fixed page preserved them by accident. Recovered carelessly, every poem becomes a paragraph. It has to be rebuilt as real verse, line by line, and there is no shortcut anywhere in it.
In every other conversion, the question is whether the formatting survived. Here the question is whether the words did. A PDF rebuild can quietly change your book, and the change will be fluent, plausible and completely invisible to any piece of software you or we could run against it.
That is why a human reads the recovered text against the original pages, and why we hand you a fidelity report at the end. Anybody can produce a file from a PDF in a minute. What takes days, and what you are actually paying for, is being able to prove that the file still says what the book said.
Optical recognition does not report doubt. It substitutes a character it is confident about and moves on. Every recognised page is proofed against its image, without exception, on every job.
Everything that exists only because there was a page must go: folios, running heads, footers, page-bound notes, page-number cross-references. Keeping them out of misplaced loyalty is how a rebuild becomes a mess.
Some things cannot come across from a fixed layout, and some should not. Each one is written into the fidelity report with the reason, so nothing is discovered by a reviewer six months later.
Twelve checks, run with the printed pages open beside the rebuilt file.
Six situations. All of them share one fact: the PDF is the only file left.
The publisher has given the book back and handed you a print PDF, which is legally generous and practically useless. There is no manuscript, no InDesign package and nobody left at the company who remembers. The book has to be recovered from the printed page, and it can be.
The interior was laid out once, the paperback sells steadily, and the eBook has never existed because the export was tried and looked appalling. It looked appalling because a fixed page cannot be poured into a reflowing screen. It has to be rebuilt.
Out of copyright, out of print, or simply out of the era before digital files. What exists is photographs of paper. Recognition and a long proofing pass will bring the text back, and the result is a book that reads like it was born this century.
You ran the file through a converter and the store refused it, or accepted it and the reviews have started mentioning the formatting. Both outcomes have the same cause. Extraction is not conversion, and the file has to be built again from the page.
A press or an estate with twenty titles surviving only as print PDFs. Rebuilt individually they will be twenty different books. Rebuilt to one standard, they are a catalogue, and the standard is what makes the twentieth cheaper than the first.
A fixed PDF cannot be enlarged, cannot be re-flowed and cannot be read aloud sensibly. Readers with low vision, and the institutions that buy for them, need a real reflowable eBook. A PDF, however tidy, is not one and never will be.
Three commitments, and the first one is the only one that really matters.
On any file involving recognition, a person reads the output against the scanned page. Not a spellcheck, not a sample. If we cannot say that, we do not deliver the file, because unproofed recognition is not text.
Everything we alter, drop or rebuild is written into the fidelity report with the reason. You will not find out from a reviewer that a paragraph went missing between the page and the file.
If any retailer refuses a file we rebuilt, on structural or validation grounds, we correct it and reissue at no charge. A recovered book has to be as clean as one born digital, and ours are.
Three positions that shape every decision in a PDF rebuild.
Extraction asks what is on the page. Rebuilding asks what the page was for. Only the second question produces a book, and everything expensive about this service follows from taking it seriously.
When the recovered file and the original disagree, the original wins and the file is corrected. Not the other way round, and never on the balance of convenience.
Some scans are too poor to be worth recovering, and some jobs cost more than the title will earn. We say so before you pay, because we would rather lose the work than deliver a book you cannot sell.
Three questions, asked of every PDF that arrives, in this exact order.
Try to select a sentence. If you can, the characters exist and can be recovered exactly. If you cannot, the pages are images and every word will have to be recognised and then proofed. This one test moves the price more than anything else.
Drawing order is not reading order. Columns, sidebars, captions, boxes and footnotes all have to be put back into the sequence a human reads them in, and that sequence exists only in the printed page and in somebody's judgement.
Folios, running heads, footers, page-bound notes, references to page numbers, hyphens created by a line ending. All of it is scaffolding for paper. All of it comes out, and taking it out cleanly is most of the craft.
Rated 4.9 out of 5 across 183 reviewed projects.
I had already paid someone else to format this book once. The chapter headings shifted every time a reader changed the font size and I only found out from a review. This team rebuilt the interior from the structure up and sent me photographs from four real devices before I uploaded anything. First launch day I have not spent refreshing the reviews in a panic.
What sold me was the audit. They came back and told me two chapters had been pasted in from another document and were carrying invisible styles, and that fixing that first would save me money later. Nobody had mentioned it in three years of self-publishing. The proof needed no corrections at all.
My book has footnotes, tables and two appendices, and every quote I got either ignored that or doubled the price because of it. Here the pricing was laid out step by step and I could see what each element cost before I committed. It arrived early and the notes link both ways, which nobody else had even offered.
The proof came back from the printer with the margins exactly where they should be, the running heads correct, and not one chapter opening on a left-hand page. This is my fourth book and the first time I have not had to pay for a second proof copy.
English is my second language and I was nervous about the back and forth. Everything was explained plainly, and when I asked the same question twice they answered as if it were the first time. The revisions were turned round inside two days and I was never made to feel like a nuisance.
I write cookbooks, so the interior is the product. Ingredients have to sit beside the method or the recipe stops working, and the last person I hired reflowed the whole thing into nonsense. These people understood the problem before I had finished describing it and the finished layout is genuinely beautiful.
Straightforward from beginning to end. A fixed price, a delivery window they actually hit, and files that went up to KDP and IngramSpark without a single rejection. I have since sent them two more titles and recommended them to my writing group without hesitation.
My memoir has photographs and I had resigned myself to them looking muddy. They optimised every image, told me honestly which two were too low-resolution to save, and suggested crops instead of shrugging. That candour is worth more to me than the discount I was chasing elsewhere.
I sent them a 480-page reference book with a nightmare of cross-references and a table on nearly every spread. They quoted it accurately, they did not come back later asking for more money, and the index still works. Any author will tell you what that is worth.
The questions authors actually ask us, answered without sales language.
PDF to EPUB conversion is the rebuilding of a fixed printed page into a flowing eBook. It is harder than any other conversion because a PDF is not a document. It is a description of where glyphs sit on a page, in a particular font, at a particular size. It does not know what a paragraph is, what order the columns are read in, or which text is a running head. A Word file, however badly formatted, still knows its own paragraphs. A PDF knows nothing, so everything has to be reconstructed.
Yes, but only by rebuilding rather than extracting, and that is a manual job. Extraction pulls glyphs out in drawing order, which on a two-column page interleaves the columns and produces sentences that look grammatical for four words and then collapse. We reconstruct the reading order by hand against the printed page, reassemble the lines into real paragraphs, strip the baked hyphens and remove the page furniture. Then somebody reads the result against the original, which is the part no software offers.
Open it and try to select a sentence with your cursor. If the text highlights, real characters are inside the file and they can be recovered exactly, with no guessing at any point. If nothing highlights, or the whole page selects as a single block, your pages are photographs of text. That means optical recognition, and it means a human proofing every recognised page against the scan. It is the single biggest factor in the price, so it is worth checking before you ask us for a quote.
When a printed book was typeset, words were broken across line endings with hyphens. In a PDF those hyphens are real characters sitting in the text, exactly as real as the letters around them. Extract the text and the hyphens come too. Reflow it to a phone screen and the line endings move, but the hyphens do not, so words are split in places they never broke. It is the clearest possible signature of a conversion nobody checked, and readers spot it within a page.
They are found and deleted, because they were drawn onto the page like any other text and extraction cannot tell them apart from your prose. Left alone, your book title and author name appear as paragraphs in the middle of the narrative, once for every page of the original, and stray digits sit alone between paragraphs where the folios used to be. Readers see a number floating in the middle of chapter four and conclude, quite reasonably, that the file is corrupted.
On clean modern print, recognition is usually above 98 percent accurate, which sounds excellent until you work out that it means several hundred wrong characters in a novel. Worse, it does not report doubt. It substitutes a character it is confident about and continues, so the errors are fluent and plausible rather than obviously broken. Every recognised page we produce is proofed against its scanned image by a person. On a scanned title that proofing is the largest single cost in the job, and it is not optional.
Yes, and it is exactly the case where automated tools fail hardest. Software reads a PDF in the order the page was drawn, which on a two-column layout means it takes the first line of the left column, then the first line of the right, and weaves them together. We unstitch the columns by hand against the printed page, rejoin any paragraph that ran from the foot of one column to the top of the next, and separate genuine sidebars from the body text before any conversion happens.
We free them from the page and reattach them to their meaning. A footnote sat at the foot of page 84 only because its marker was on page 84. Take away the page and the note is just a small orphaned block of text with no relationship to anything. We match each note back to the marker it belongs to and rebuild the pair as a bidirectional link, so a reader taps the marker, reads the note, and returns to precisely the sentence they left.
The data will, if somebody rebuilds it. The table will not, because there is no table. In a PDF a table is a set of drawn lines with words floating in the gaps between them, and extraction produces those words in the order they were drawn, which is usually a jumble. We reconstruct each one as a real table with proper header cells, and where a wide printed table cannot survive a six-inch screen we restructure it into a stacked, labelled form that a phone reader can actually use.
It is normal, and it is more information rather than less. Crop marks, registration marks, colour bars and bleed all tell us about the printed page, and none of them belong in an eBook, so they are removed along with the rest of the print furniture. The images inside a print-ready PDF are usually held at print resolution in a print colour space, which means they are heavy and wrongly coloured for a screen. Both of those are corrected during the rebuild.
Usually, and it is some of the most rewarding work we do. Foxed paper, show-through from the reverse of the sheet, broken type and a tight gutter all make recognition struggle, so the human proofing pass grows and the price grows with it. Send us a handful of representative page images before you commit to anything. We will tell you honestly whether the scan can carry a book, and if the answer is no, we will say so rather than take the money.
You can upload it, and Amazon will convert it for you, badly, using exactly the extraction that produces everything described on this page. Your running heads will appear in the prose, your hyphens will surface mid-word and your columns will interleave. The result is published under your name, and the first people to notice are the readers who paid for it. A PDF is a print artefact. It is not an eBook, and no store will turn it into one on your behalf.
You can, and for a legacy title it is often the most valuable thing in the delivery. The recovered book is handed back as a clean, styled Word document with real heading styles, which for many authors is the first time in years that an editable version of their own book has existed anywhere. It is what a corrected second edition starts from, what a new print run is set from, and what makes any future work on the title cheap rather than catastrophic.
Nothing at all, unless it is rebuilt. A printed index is a list of page numbers, and a reflowable eBook has no pages, so every entry points at something that does not exist. There are two honest answers. We can rebuild the index as live links that land on the paragraph each entry refers to, which is what most readers actually want. Or we can carry the print pagination across as page-list markers, which is what libraries and academic citation require. Some books want both.
Because there is more work in it and less of it can be automated. Every other source format still contains a document. A PDF contains a page. The text has to be recovered, the reading order reconstructed, the paragraphs reassembled from individual lines, the hyphens stripped, the furniture removed and the notes reattached, and on a scan every word has to be proofed by eye first. It is the only conversion where a person reads the entire book twice, and that is what the price reflects.
The character of it, yes. The geometry, no, and this is worth being clear about because it is where most disappointment comes from. Your typeface, your chapter opener treatment, your ornaments and your general typographic voice can all carry across. What cannot is anything that depended on a page of a fixed size: the exact position of a drop cap, a photograph bled off the corner, a line that fell perfectly at the foot of a recto. The reader now controls the page, and there is no way to take that back.
Then it is the most fragile thing on this page and it needs saying up front, because recovering it carelessly is destructive. Poetry survives in a PDF by accident, since the fixed page happened to preserve the line breaks the poet intended. Extract it and every poem becomes a paragraph, or worse, a paragraph with hyphens in it. We rebuild verse as real line-level markup, with hanging indents for runover lines that hold at any type size, and we do it line by line.
Yes, and it happens more often than you would expect with scans of older books. Missing, duplicated and out-of-sequence pages are found during the read-back against the original, because someone is going through the whole book in order. We will tell you exactly which pages are absent and ask you to supply them. What we will not do is quietly bridge a gap and hand you a book with three pages missing from chapter nine.
Longer, and the calculator shows honest windows for both. A digital PDF of a straightforward novel takes somewhat longer than the same book as a .docx, because the reading order and the paragraphs have to be rebuilt. A scanned academic title with hundreds of footnotes takes considerably longer, because the recognised text has to be proofed page by page before any structure can be trusted. We give you a fixed date after opening the file, not before.
Yes, and it will save you money. An InDesign package still contains a document: real paragraph styles, real text threads, real anchored objects. A great deal of the reconstruction we would otherwise be doing by hand is already recorded in it. If the InDesign file exists anywhere, on any drive, send it. The same goes for an old Word manuscript. The PDF should be the source of last resort, and it is only because it so often is the last resort that this service exists.
Digital or scanned, one column or two, clean or foxed. We will open it, tell you whether the text can be recovered and what it will take, and quote a fixed price and a real date before you commit to anything.