An ad campaign ends. An edition doesn't.
Somewhere there is a reader who would have loved this book more than most of the people who have read it.
They read constantly. They have been looking for exactly this argument, or this story, for years. They will never read a word of it, for no reason at all except that the two of you learned different languages as children.
You didn't skip it. You couldn't see the end of it.
Translation is priced as a project, and a project with no visible ceiling doesn't get started.
That's the whole reason it sits undone in almost every writer's life. Not parochialism, not lack of ambition. You went to find out what it would cost, the answer was a conversation, and the conversation was going to be about your budget. So it stayed on the list, and the list is where it still is.
Every other thing you spend on is consumed. This one is owned.
Here is the comparison writers make, and it's the wrong one.
Translation gets weighed against marketing, because both are money spent to reach more readers. But they are different kinds of object. Advertising is consumption: it works while the money is going out and stops when it stops. A campaign has a date and then it's over, and what you have afterwards is whatever it did while it ran.
An edition is capital. It sits on a store, in a language, and keeps selling with no further input for as long as the book exists. Nobody has to keep paying for it to keep working.
It's the only spend in publishing that produces a second asset rather than a temporary increase in attention on the first one. And it compounds in a way nothing in the drafting stage can: the manuscript can only be made better, once. An edition makes another of it.
Your book doesn't contain the answers
A translation cannot be a copy, because your manuscript does not hold enough information to make one.
Take a language with a formal and an informal you. Your narrator addresses the reader — and now somebody has to choose. That choice sets whether your narrator is intimate or holding the reader at arm's length, across every sentence of the book. You never made that decision. English never asked you to. The information isn't in your manuscript and no amount of care extracts it, because it was never there.
Multiply that by several thousand. Where the register sits. Whether a joke that lives in word order becomes a different joke or gets abandoned. Whether the term you coined is borrowed, calqued or replaced. Whether a sentence that is deliberately ungrammatical stays ungrammatical, in a grammar with entirely different rules to break.
Every one of those is a decision your book didn't make. Which means a translation is not your book restated. It is your book plus several thousand judgements taken by somebody else, in service of what you meant.
That isn't a compromise. It's what translation is, and it's why the result is a second original rather than a copy — sold beside the first, not instead of it.
The machine is flawless on the sentences that don't matter
Machine translation is genuinely good now, and it's good in a specific and misleading way.
It is near perfect on sentences that carry only information. The train leaves at six. The company was founded in 1974. Those have a correct answer and it finds it.
It fails on the sentences that carry you. Rhythm. The register you settled after three drafts. The joke that lives in word order. The repetition that is deliberate and reads as an error. The rule you broke on page one so that breaking it on page two hundred would land.
So you get a perfect translation of the parts of your book anybody could have written, and an approximation of the parts only you could. It's the same thing you're afraid of in a proofreader — being smoothed into a register that belongs to nobody — except here you can't read the result to find out it happened.
"It won't sound like me"
Half of that is true, and the honest answer says so.
A translation is a performance of your book by somebody else, the way a recording is a performance of a score. There is no version of this where it arrives with your fingerprints on it and none of theirs.
What follows isn't resignation. It's a change in what you're actually asking for. You are not commissioning a copy to be checked line by line against the original — that thing doesn't exist. You're commissioning an edition, and the only question that matters is whether the person making it understood what the book was doing.
Which is why it stands as its own edition, beside the original. Still your work, under your terms.
What it costs
$30 buys thirty minutes of our team's time. In one block: the length and the language read, and the number of blocks named before anything starts. The book itself takes more than one, and you'll know how many in advance rather than afterwards.
Thirty minutes buys our hands. A translator's work is a translator's work — passed through at cost, with the number in front of you before it runs.
You can do all of this yourself, here, for nothing.
The reader at the top of this page is not hypothetical. They exist right now, in a language you don't write in, and the only thing standing between them and your book is an edition nobody has made yet.