You are not starting from a blank page. You have a Word file with fifteen years of drafting inside it, a print PDF, an InDesign package nobody can open, or a back catalogue trapped in a format Amazon stopped caring about. Conversion is not a button. It is forensic work: keeping what the file got right and rebuilding by hand every single thing the software quietly threw away.
Conversion is the migration of a book from a format that suited its old life into one that suits its next one. The mistake almost everybody makes is thinking of it as a translation, where the words move and the meaning survives. It is closer to a house move. Things get broken in transit, and somebody has to notice which ones.
Automated tools are very good at the ninety percent of a document that is ordinary prose. They are catastrophically bad at the remaining ten percent, which is where all your character lives: the tables, the footnotes, the pull quotes, the poem set in italics, the chapter number that was a picture rather than text, the small caps that were faked with a font swap.
Our eBook conversion services put a person between the old file and the new one. The tool does the bulk work. A human then goes through what came out, compares it against the original, finds the things that silently vanished, and rebuilds them properly in the target format.
The dangerous damage from an automated conversion is never the damage you can see. A file that refuses to open gets fixed immediately, because it is obvious. A file that opens beautifully and has quietly dropped the second half of every footnote gets published, and you find out from a review.
So the first thing we do with any conversion is not convert anything. We read the source and write down what is in it, so that afterwards there is something to check the output against.
A heading that was a heading becomes a paragraph that happens to be bold and large. The contents list built from it is then empty, and no error is reported anywhere.
Footnote markers survive as superscript text and the links behind them disappear. The reader can jump to the note and then has no way back to the sentence.
Some pipelines rasterise anything they cannot lay out. Your data is now an image, unsearchable, unreadable at any other size and invisible to a screen reader.
The tool is the cheap part. Everything below is the part you are actually paying for.
Before anything is converted we catalogue the original: how many heading levels, how many notes, how many tables, which fonts, which styles are real and which are faked with direct formatting. That inventory becomes the checklist the output is graded against, item by item.
Wherever the source relied on visual formatting instead of real styles, the converter has nothing to work from and invents something. We throw that away and tag the element for what it actually is, so the chapter is a chapter and the sidebar is a sidebar.
Tables that were rasterised are rebuilt as live tables. Footnotes that lost their links are re-linked in both directions. Cross-references that pointed at page numbers are repointed at the thing they were referring to.
Images extracted from a PDF or a legacy file are usually downsampled, colour-shifted and stripped of their alt text. We pull the highest-quality version available, re-export it for modern screens and describe it properly.
When several titles are converted together, they are converted to one documented standard. Book one and book nine end up with the same chapter openers, the same margins, the same ornaments and the same back matter.
The finished file is read side by side with the source. Every item on the autopsy checklist is ticked or explained. Nothing is declared finished because a progress bar reached the end.
Six migrations, each one arriving as somebody else's problem file.
The economics of conversion change completely with volume, and so does the method. Deciding which of these you are doing, before anyone opens a file, saves a great deal of money.
One book, one source file, one target. The whole job is bespoke: we treat the file on its own terms, make individual judgements about every table and every image, and hand it back as a finished EPUB 3.
This is right when the book is standalone, when the source is unusual, or when the title matters enough that you want somebody thinking about it rather than processing it.
Five, twenty or ninety titles moving together. Here the first book is the expensive one, because that is where we build and document the house standard. Every title after it is measured against that standard instead of being reinvented.
Per-title cost falls sharply with volume, and the real prize is consistency. A backlist that looks like one publisher made it sells better than the same books looking like nine different accidents.
Six failures that appear in almost every automated conversion we are handed. Not one of them raises an error message.
Different origins fail in different ways. These are the eight we meet most, and the specific treatment each one gets.
| Element | How we build it | Why it matters |
|---|---|---|
| Word or Docs | Styles mapped to semantic elements, direct formatting stripped, manual breaks and tab indents removed before conversion runs. | A Word file that looks tidy is usually held together by direct formatting, which converts into thousands of unusable inline rules. |
| PDF, digital origin | Text extracted with its reading order verified, line-end hyphens removed, columns unstitched and the flow reassembled paragraph by paragraph. | PDF stores where glyphs sit, not what they mean. Reading order is frequently the order things were drawn, which is not the order they are read. |
| PDF, scanned | Optical recognition first, then a human proof against the scan, then structure applied. Priced as a rebuild rather than a conversion. | There is no text in a scan, only pictures of text. Recognition introduces errors that only a person comparing pages will ever catch. |
| InDesign | Paragraph and character styles mapped by hand, anchored objects re-anchored, absolute positioning discarded and replaced with flow. | The built-in export produces a file that validates and reads badly, because it tries to preserve a page geometry the reader has already thrown away. |
| EPUB 2 | Deprecated markup replaced, nav document generated, NCX retained as a fallback, metadata brought up to current spec. | Retailers now reject constructs they accepted a decade ago. The file that sold fine in 2013 will not ingest today. |
| MOBI or PRC | Unpacked, text and structure recovered, then rebuilt as EPUB 3 rather than converted directly across. | MOBI is a dead end. Converting out of it faithfully means inheriting its limitations, so we rebuild from the recovered content instead. |
| HTML or web content | Presentational markup removed, semantic tags applied, remote assets localised into the package. | Remote images and web fonts are a validation failure in EPUB. Everything the book needs must live inside the file itself. |
| Plain text or RTF | Structure inferred, confirmed with you, then applied. Chapters, breaks and emphasis are reconstructed from scratch. | A plain text file has no formatting to preserve, so nothing is lost. It also has nothing to work from, so everything is a decision. |
Verification is the whole service. Anyone can run a file through a tool. Six checks decide whether what came out is still your book.
The output is read against the source, page by page for a short book and chapter by chapter for a long one. A person is looking for missing text, dropped emphasis, and paragraphs that have merged into their neighbour.
Headings, images, notes, tables and list items are counted in the source and counted in the output. If the numbers do not match, we know exactly how many things went missing before anyone has to hunt for them.
The extracted text of both files is compared automatically to surface substitutions the eye slides over: broken ligatures, mangled accents, en dashes turned into hyphens, quotation marks reversed.
The converted file is opened on e-ink, on tablet and on a phone. Conversion artefacts love the extremes: they show up as strange spacing on a small screen or an image that refuses to render on old firmware.
EPUBCheck runs against the final package, and the file is put through the retailers' own ingestion checks. Legacy conversions fail here most often, and it is far better that they fail here than in front of you.
On a multi-title job, all converted books are compared to one another. Chapter openers, margins, ornaments and back matter must be identical, because a reader moving from book two to book three will notice if they are not.
The converted book, plus the evidence that the conversion did not quietly cost you anything.
The migrated book as a valid EPUB 3 package, with real structure underneath it rather than the flattened remains of the original formatting. This is the file the stores want and the file you should keep.
A derived file confirmed in Kindle Previewer, which matters more on a conversion than on a fresh build because legacy artefacts tend to reveal themselves specifically on Amazon's rendering path.
A plain list of everything found in the source, everything carried over, and anything we deliberately changed, with the reason. This is the document that tells you the migration was honest.
Where the source was Word or InDesign, you also get the tidied version: real styles, no direct formatting, no phantom breaks. Useful for the next edition and for any print work.
Every image pulled out of the original at the best quality it could yield, re-exported for modern screens and named consistently, so you never have to go back into a dead file to find a picture.
The XHTML, the CSS and the package document. On a back-catalogue job this is what makes the next twenty titles cheap, because the standard is written down rather than living in someone's head.
Five stages. The first one is the one nobody else does, and it is why the last one is boring.
We open the source and inventory it. What is genuinely styled, what is faked, how many notes, how many tables, which fonts, what is going to break. You get that list, and it doubles as the quote.
Before any conversion runs, the original is cleaned: direct formatting stripped, phantom breaks removed, styles normalised. A conversion is only ever as good as the file fed into it.
The tool does the bulk work. Then the rebuilding starts, which is most of the job: structure retagged, notes re-linked, tables reconstructed, images re-exported, CSS written from scratch.
Counts reconciled, text diffed, output read beside the original. Anything missing goes back into the rebuild until the two files agree.
EPUBCheck, hardware, ingestion checks, then delivery with the loss report. On a bulk job, the whole set is swept for consistency before any of it ships.
Conversions are priced on the length of the book and the state of the source file. A tidy Word document is cheap. A scanned print PDF is a rebuild with a conversion attached to it.
Required steps are marked with *. Optional extras can be removed from the summary with the cross icon.
What kind of book is being migrated. This decides how much of the original structure has to be reconstructed by hand.
The approximate word count of the existing book. If you only have a page count, multiply by 300 and we will confirm from the file.
What you are converting from. This is the single biggest factor in a conversion quote.
How much non-prose material is inside. Notes, tables and cross-references are where conversions break.
How many images are inside the book. Images extracted from PDFs and legacy files usually need recovery work.
Standard runs at the pace verification actually needs. Faster tiers add hours, not shortcuts.
Two rounds are included. Most conversion revisions are about taste, because the errors are already gone.
Conversion covers structure and format only. We do not edit, rewrite, proofread or fact-check the text we migrate. If the source has a typo in chapter three, the converted file will have the same typo in chapter three, faithfully preserved.
Both produce a file. Only one of them produces the same book you started with.
| What you are comparing | Automated Batch Conversion | Book Formatting Experts |
|---|---|---|
| What happens to structure | ✕ Guessed from appearance. A bold line becomes a heading, or does not, depending on the font size that day. | ✓ Read from the source, confirmed with you where ambiguous, then tagged deliberately. |
| Missing content | ✕ Nobody looks. Whatever the tool dropped is gone, and the first person to notice is a reader. | ✓ Counted before and after. If the source had 214 footnotes, the output has 214 footnotes. |
| PDF handling | ✕ Text pulled in drawing order with hyphens baked in and columns interleaved into nonsense. | ✓ Reading order reconstructed, hyphens stripped, columns unstitched, then proofed by a person. |
| A back catalogue | ✕ Each title converted in isolation, so nine books end up with nine different chapter styles. | ✓ One documented standard applied across the set, and a consistency sweep before anything ships. |
| Legacy files | ✕ Converted faithfully, which means the old file's limitations are carefully preserved in the new one. | ✓ Content recovered, then rebuilt to current spec, so the book stops being a decade old. |
| Evidence | ✕ A file and a shrug. | ✓ A loss report listing what was in the source, what came across, and what we changed on purpose. |
Six recurring jobs, each with a characteristic way of going wrong.
Out of print, the publisher has reverted rights, and all that exists is a print PDF or a scan. The text has to be recovered, proofed against the original and rebuilt from nothing. It is the most rewarding conversion there is, because the book genuinely comes back.
Full of boxed callouts, frameworks, tables and diagrams, all of which the export flattened into identical grey paragraphs. Rebuilding it means deciding what every box was for and giving each kind its own identity again.
Hundreds of footnotes, a bibliography, cross-references and an index of dead page numbers. Nothing here converts cleanly. Everything has to be re-linked, and every link has to be tested in both directions.
The single most fragile thing to convert. Line breaks become paragraph breaks, hanging indents disappear, spacing that carried meaning is normalised away. It is rebuilt as real verse markup, line by line, and there is no shortcut.
Exported from a page layout program where every element sat at a coordinate. Ingredient lists and method steps have to be re-associated with their recipe, and the photography has to be re-extracted at usable quality.
Twenty or ninety titles, converted at different times by different people, none of them matching. The job is not really conversion at all. It is imposing one standard on a decade of inconsistency.
Anyone can run a file through a converter. The tool costs nothing and takes a minute. What you are buying is the person who then reads the result against the original and finds the fourteen things the tool did not mention it had thrown away.
That is why our conversions come with a loss report. If we cannot show you what survived and what did not, we have not really converted anything. We have just moved a file and hoped.
You cannot notice a missing footnote unless you counted the footnotes first. The autopsy exists so that the verification has something to check against.
Faithfully preserving a legacy file's constraints is not fidelity, it is laziness. We recover the content and rebuild it to current spec, so the book is fit for the next decade rather than the last one.
Some things cannot be carried across, and some things should not be. Every one of those is written down in the report with the reason, so nothing is discovered later.
Twelve checks run on every converted file before it is allowed out.
Six situations. If any of them is yours, you have a conversion problem rather than a formatting one.
Something exists. It might even be selling. But the contents list is empty, the notes go nowhere, or the reviews mention formatting. Starting again from the manuscript is not necessary. The book inside the file is fine and only the structure has failed.
The publisher has handed the book back and given you a print PDF, or nothing at all. The text has to be recovered from whatever survives, proofed, and rebuilt into a file you own outright and can sell yourself.
Titles converted years ago are starting to be refused at ingestion, or they render badly on current hardware. The market did not change, the specification did, and the files are being left behind.
Everything went to InDesign, the layout is beautiful, and there is no eBook because the export was tried once and looked terrible. That export was never going to work. The file needs a person, not another attempt at the button.
You have been in one ecosystem and want out, or you want in somewhere new, and the file you have will not ingest there. Conversion is what gets your book from the format it is trapped in to the one that sells.
Dozens of titles, converted by different hands over years, no two of them alike. What you need is not twenty conversions. You need one standard, and then twenty titles brought into line with it.
Three commitments, each one specific enough that you would notice if we broke it.
Everything found in the source is either in the output or listed in the loss report with a reason. You will never discover a missing chapter six months after launch, because we will have told you.
If a store refuses a file we converted, on any structural or validation grounds, we correct it and reissue at no cost. Getting the file accepted is the job, not an optional extra to it.
We migrate structure, never wording. Not a comma is changed without you asking for it. A conversion that quietly improves your prose is a conversion you can no longer trust.
Three habits that separate a conversion from a file transfer.
The tool always reports success. It reports success when it has flattened your headings and eaten half your notes. Treating every conversion as guilty until verified is the only workable position.
Not the converted file, not the tidy-looking preview. When the two disagree about what the book contains, the source wins and the output gets rebuilt.
On any multi-title job, the first book is where all the arguments happen. Once the standard is written down, the rest of the catalogue stops being expensive and starts being fast.
Three questions, asked in this order, on every conversion we take.
Not what the file looks like. What it contains. Real styles against faked ones, live notes against superscript text, true tables against pictures of tables. This is the inventory everything else depends on.
Some things cannot cross: fixed page numbers, running heads, coordinates, fonts that cannot be embedded. Naming them up front turns them into decisions rather than into surprises.
Whatever is left. This is the real scope of the job, it is where the cost lives, and it is what we quote on. Everything else is the tool doing the easy ninety percent.
Rated 4.9 out of 5 across 232 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.
eBook conversion services migrate a book that already exists in one format into a working eBook in another. You need one when the file you have is the wrong shape: a print PDF with no eBook version, an InDesign layout that will not export usefully, a Word document that produced a broken upload, or a legacy MOBI that stores no longer accept. If you are starting from a finished manuscript with nothing built yet, that is a formatting build rather than a conversion, and it is priced differently.
Everything that is not ordinary prose. Headings lose their semantic tag and become bold paragraphs, so navigation comes out empty. Footnote return links vanish. Tables are sometimes rasterised into images, making your data unsearchable and invisible to screen readers. Small caps and italics that were faked with a font swap disappear entirely when the font does. None of this raises an error. The tool reports success, the file opens, and the damage is only found by someone comparing the output against the original.
Yes, but it is the hardest conversion we do and it is priced accordingly. A PDF records where each glyph sits on a page, not what any of it means. Reading order is often the order things were drawn, so a two-column layout interleaves into nonsense. Line-end hyphens are baked in as real characters and reappear mid-sentence after reflow. We reconstruct the reading order, unstitch the columns, strip the hyphens and then proof the result against the original by eye.
It very rarely is. The export tries to preserve a page geometry that a reflowing screen has already discarded, so blocks arrive pinned to coordinates that mean nothing and text lands on top of other text. Anchored objects drift to the wrong paragraph. Character styles come through as anonymous spans. We map your paragraph and character styles by hand, re-anchor the objects, throw the positioning away and rebuild the book as flow rather than as pages.
Yes, and we rebuild rather than convert directly. Converting a MOBI faithfully means carefully preserving all the limitations of a format that has been retired, which helps nobody. Instead we unpack the file, recover the text and whatever structure survives, and then build a current EPUB 3 from that recovered content. The result validates, ingests at every store, and will still work in ten years, which the original will not.
Less per title than a single book, and it falls further with volume. The first title is the expensive one, because that is where we build and document the house standard: chapter openers, margins, ornaments, back matter, metadata. Every book after that is measured against a standard that already exists rather than being reinvented. Send us the list of titles and the source formats and we will price the whole migration as a project rather than as a stack of separate jobs.
That is one of the most common repairs we do. The blank screens are usually empty paragraphs that the original author used to push text down a page, and which the converter dutifully preserved. The stray fonts are typically inline font declarations left behind by a word processor. Both are removed, the structure underneath is retagged properly, and the stylesheet is rewritten from scratch so the file has a small, readable set of rules instead of thousands of contradictory ones.
The text of them, yes. Their usefulness, only if we rebuild it. An index in a converted file is a list of dead page numbers pointing at pages that no longer exist, because an eBook has no pages. We can rebuild it as live links that land on the paragraph each entry actually refers to. The same is true of cross-references. Left alone, they say see page 214, which is meaningless. Rebuilt, they take the reader there.
We handle the whole thing, recognition included. A scan contains pictures of text and no text at all, so optical recognition has to run before anything can be converted. Recognition is confidently wrong in ways that are easy to miss: it swaps letters, mangles accents, and invents plausible words. So a person proofs the recognised text against the scan before structure is applied. It is priced as a rebuild rather than a conversion, because that is honestly what it is.
Yes, and we build them all from one verified source rather than converting sideways from one output to another. That means a validated EPUB 3, a Kindle-compatible file confirmed in Previewer, and where useful a clean Word master and an archive copy. Chaining conversions is how errors compound: each pass inherits the last one's damage and adds its own. One source, several derived outputs, all checked separately.
By writing the standard down before converting anything, and then converting against it. Chapter opener treatment, type scale, margins, ornaments, back matter order and metadata conventions are all fixed in a documented style, and each title is built to it. Before delivery, the whole set is swept side by side to catch drift. A reader moving from book two to book three in a single evening will absolutely notice if the chapter headings have changed size, even if they cannot say why it bothers them.
It is normal, and it is exactly what we clean before conversion runs. Manual breaks and tab-based indents are instructions for a printed page, and an eBook has no page for them to act on. Left in, they produce screens of blank space and indents that appear in the middle of sentences after reflow. We strip them, apply real styles that mean the same thing structurally, and only then convert. A conversion is only ever as good as the file fed into it.
Yes, and it is a common request when a new edition is due or a print interior is being planned. You get a properly styled Word document with real heading styles rather than a flat dump of text, so the structure you paid for in the EPUB survives into the document. It is worth doing this at the same time as the eBook conversion, because both are built from the same verified source and stay in step with each other.
Internal links are re-pointed at the correct target in the new file, because the underlying document structure has changed and the old destinations no longer exist. External links are carried across and then tested. Dead ones are reported to you rather than quietly shipped. Legacy files in particular tend to be full of links to sites that closed years ago, and a reader tapping one and reaching an error page reflects on your book, not on the internet.
Longer than one title and much less than ten times as long. The first book carries the standard-setting work and takes a normal single-title timeline. After that, the titles move in parallel and each one is faster than the last. Ten typical novels from clean Word sources is a matter of a few weeks. Ten scanned academic monographs is a different conversation entirely, and we will give you a real schedule after we have opened the files.
No, and nothing you buy will, because a reflowable eBook has no fixed pages. The number of screens depends entirely on the device and the type size the reader has chosen. What we can do is embed page-list markers that map to the print edition, which is what libraries and academic citation need. If that matters to your readers, say so at the start and we will carry the print pagination across as reference points.
That is one of the things we are asked for most. Forward whatever the retailer told you, together with the file they refused. The message tells us which class of failure the ingestion pipeline hit, and the file tells us why. Most rejections come from a small set of causes: deprecated markup in a legacy file, remote resources referenced from inside the package, a broken manifest, or an image the file claims to contain and does not. All of them are fixable.
Never. We migrate structure and format, and we do not touch a single word without you asking. If your source has a typo, the converted file will have the same typo, preserved exactly. This is deliberate. A conversion that silently improves your prose is a conversion you cannot audit, and you would have no way of knowing what else it decided to help with. If you want the text corrected, that is editing, and it is a separate job.
Usually, and often more cheaply than converting from scratch, because the text has already been extracted. We open what you were given, inventory it against your original source if you still have it, and rebuild the parts that failed. Occasionally the file we are handed is beyond repair and going back to the source is genuinely cheaper. We will tell you which it is after looking, before you commit to anything.
Word, Google Docs, RTF, plain text, HTML, InDesign and IDML packages, print-ready and digital PDFs, scanned page images, EPUB 2, EPUB 3 that needs rebuilding, MOBI, PRC and AZW. If your book exists in some other format entirely, send it anyway. We have opened stranger things than you would expect, and the answer is almost never that the content cannot be recovered. It is only ever a question of how much hand rebuilding sits between the source and a working file.
Upload whatever you have, however bad it is. We will open it, inventory what is really inside, and come back with a written scope, a fixed price and a date. If the file is fine and only needs a small repair, we will say that too.