Probability of a book being published

[Aiwendil]

I was just contemplating what the chances are, given that you have a finished draft that you attempt to traditionally publish (but remaining agnostic about the quality, marketability, or any other attributes of that draft), that that draft will end up published, just going by the numbers. I'm far, far from an expert on any of this, but I just wanted to share my thought process in case anyone else finds it interesting.

Of course, hard, fast numbers are not available, but estimates for the number of queries an agent receives seem to vary from something like 100 per week (5200 per year) to 100 per day (36,500 per year), with most estimates falling somewhere around the 10,000-15,000 per year range. Of course, this includes queries from people who have not, in fact, completed a manuscript, so the number of actually valid submissions will be a little lower than the raw number. So let's say 10,000 per year.

Most estimates that I've found for the number of clients an agent signs per year are something on the order of 2-4. So let's estimate 3 as an average. That means the probability of a given agent signing an author for a particular book is 3/10,000 = 0.0003, or 0.03%. Very low! But of course, you don't submit to a single agent. Assuming no correlations between agents' response probabilities (which is probably true to a decent approximation), your chances of getting all rejections are (1-0.0003)^N, where N is the number of agents you submit to. Let's say you submit to 100 agents (definitely on the high end, but we're imagining that one is going all out on trying to get this published). Then the probability of getting all rejections is (1-0.0003)^100, which comes out to about 97% - in other words, there is a 3% chance of getting at least one offer of representation.

Estimates of how many agented books are accepted by a publisher seem to be all over the place. This article (https://jerichowriters.com/if-an-agent-accepts-your-work-what-are-chances-of-getting-published-2/) first claims that a good agent sells 2 out of every 3 books (67%) they take on submission. But then it considers things from the publisher's end, where only about 1% of books submitted get published, which means that if an agent submits to about 10 publishers, there is a (1-0.01)^10, or about 10% chance that a given agented book will be published. Let's split the difference and assume about a 35% chance that a book represented by an agent will find a publisher. That leaves us with a final probability of 0.03*0.35 = 0.0105, or just a tad over a 1% probability that a completed book will end up traditionally published. Not great, but not quite as improbable as winning the lottery!

I'm curious whether anyone else has tried calculating this number, or if there are places where my estimates seem off or could be improved.

 

[Helix]

One of the problems with calculations like this is they're only meaningful if every manuscript is publishable. A lot of them aren't.

 

[Aiwendil]

Yes, that's why, as I said, this is agnostic about the quality or publishability of the book. But presumably, that's the number that's actually relevant, since there's no objective way of determining whether the book you've finished is publishable (without querying it and finding out - and presumably if a person is querying a book, that person at least believes that it's publishable). So there isn't actually any further information there that you can use.

 

[Fuchsia Groan]

One of the problems with calculations like this is they're only meaningful if every manuscript is publishable. A lot of them aren't.

That’s the thing. For some manuscripts, the odds will be way better than this calculation suggests. For others, they will be way worse. There’s no way to get good, hard and fast numbers on a particular book. I’ve been through it when I was on sub to publishers, and in the end it doesn’t really help to guess at the likelihood of a sale. Either it happens or it doesn’t.

That’s just my experience, though! I’ve had one book that died on sub and three published (or soon to be). The more I learned about the market, the more I could tell when I was getting “warmer,” getting promising responses that suggested something might sell. When I started out, I knew nothing of the market and didn’t care about the market, and that showed in my abysmal results. So even for one writer with a constant ability level, the success rate can vary wildly.

Not very helpful, I know. I just think learning about the market is the most important thing any writer can do, whether they want to be trade or self-published. Even if you’re never really going to “write to market,” finding the overlap between your writing desires and the market is a major determinant of success.

 

[Helix]

Yes, that's why, as I said, this is agnostic about the quality or publishability of the book. But presumably, that's the number that's actually relevant, since there's no objective way of determining whether the book you've finished is publishable (without querying it and finding out - and presumably if a person is querying a book, that person at least believes that it's publishable). So there isn't actually any further information there that you can use.

But to get meaningful stats, you can't be 'agnostic' about book quality. You also can't assume that someone will keep querying if they get an acceptance. You can't assume that everyone goes through an agent. You can't assume that a debut author will have the same chance as a mid-lister or that either of them will be in the same boat as field leaders.

 

[Nether]

While I imagine most people acknowledge a lot of it comes down to luck, somehow I don't think you can predict odds like that. There are a crazy number of variables and some can get hyper-specific.

But presumably, that's the number that's actually relevant, since there's no objective way of determining whether the book you've finished is publishable (without querying it and finding out - and presumably if a person is querying a book, that person at least believes that it's publishable).

Technically not even then, because a terrible query can sink even the best manuscript.

Also keep in mind a portion of queries agents receive are for manuscripts that haven't even been written yet, because some people try to bounce ideas off agents.

 

[mccardey]

I don't think it's possible to calculate - there are just too many variables. That's why I feel its better for people to enjoy the act of writing rather than pinning their hopes on the idea of getting published. Even "getting published" means very little - it's no guarantee of anything, and it can be a fairly shattering experience when things go wrong. And things do go wrong.

Better to write because you can't not write, I think.

 

[Unimportant]

Probability of a bad book being published? Near but not totally zero.

Probability of a phenomenally faboo book being published? Somewhere between a hundred and zero but definitely leaning towards the hundred.

Probability of an excellent to good to adequate book being published? In the lap of the gods, all genre imperatives and individual drives and luck on the day and grabbiness being equal, which it's never not. I

 

[mccardey]

all genre imperatives and individual drives and luck on the day and grabbiness being equal, which it's never not. I

I feel you had more to say...

 

[Unimportant]

I would but I am busy making blackberry coulis. Apologies to all.

 

[ChaseJxyz]

This is like trying to come up with a "real" result from the Drake equation. There are so, so many variables that you cannot possibly know or account for.

I work in marketing, and one of the companies made products for gamers. Part of our efforts was doing brand integrations with different games. One year we got a really big title, for a really big IP, and we were all very excited! Everything was projected to go great. The game was about being cops shooting bad guys, and a few weeks before the game came out, a George Floyd-esque situation happened and was all over the news. No one wanted to play a game about being a cop shooting a bunch of people, and no news outlet wanted to give this game good coverage. And also we looked like huge jerks for doing this brand thing, even though contracts were signed over a year ago and we've been planning this whole thing for a long time.

Charles Stross had to throw out several novel ideas recently because Real Life made them unsellable. No one wanted to read a story about a global pandemic, because we're all living it and anything written awhile ago would look unrealistic based on our lived experiences. And I'm sure every writer who had a manuscript gathering dust about high school girls kissing angels or demons or whatever thanked their lucky stars when Twilight got big. Maybe everything is going great, you're gonna get your 7-figure advance, then there's a merger and McPenguin House Hill has decided that your book no longer fits their roadmap, so, uh, sorry! Sucks to be you.

I love numbers. But this isn't like hunting for a shiny Pokemon where you will PROBABLY get it if you grind and grind and put a bunch of hours into it, there aren't tricks that are guaranteed to increase your odds. There are things you can do to make your work more publishable (studying trends, having the ideal length, including specific themes or subjects) but those are vague "it'll help" and not "this will double your chances." You also have to consider how fucking depressing it is to write to min-max numbers instead of doing it to create something artistic or that you genuinely enjoy. I've done a lot of writing for SEO, which is all about numbers, and it sucks so hard. There is no creativity, you're forced to create content that Just Isnt Good for a human to read, but The Algorithm likes it. Is that something you want to do for whole books? This IS a feasible strategy for self-publishing where you can hyper-target certain markets on Kindle, but you said trade publishing, where this is not possible.

 

[Helix]

ChaseJxyz's mention of the Drake equation reminded me of all those silly equations on the probability of life arising on Earth.

It's 1. The probability of life arising on Earth is 1.

Anyway, I return you to your normal programming.

 

[mccardey]

I've done a lot of writing for SEO, which is all about numbers, and it sucks so hard. There is no creativity, you're forced to create content that Just Isnt Good for a human to read, but The Algorithm likes it. Is that something you want to do for whole books? This IS a feasible strategy for self-publishing where you can hyper-target certain markets on Kindle, but you said trade publishing, where this is not possible.

I've done a shedload of writing TVCs. It paid really well (not now - but it used to) and every job had a guaranteed outcome. But it wasn't like writing.

 

[Unimportant]

What is a TVC?

 

[mccardey]

What is a TVC?

Television commercial. Prolly before your time.


SALE ENDS THURSDAY!!

 

[lizmonster]

If you scatterplot every author in the world and throw out the outliers, nobody gets published.

This is not a problem to which it is meaningful to apply math.

Also, is it time for a Slushkiller link? I think it is.

 

[stephenf]

I was just contemplating what the chances are, given that you have a finished draft that you attempt to traditionally publish (but remaining agnostic about the quality, marketability, or any other attributes of that draft), that that draft will end up published, just going by the numbers. I'm far, far from an expert on any of this, but I just wanted to share my thought process in case anyone else finds it interesting.

Of course, hard, fast numbers are not available, but estimates for the number of queries an agent receives seem to vary from something like 100 per week (5200 per year) to 100 per day (36,500 per year), with most estimates falling somewhere around the 10,000-15,000 per year range. Of course, this includes queries from people who have not, in fact, completed a manuscript, so the number of actually valid submissions will be a little lower than the raw number. So let's say 10,000 per year.

Most estimates that I've found for the number of clients an agent signs per year are something on the order of 2-4. So let's estimate 3 as an average. That means the probability of a given agent signing an author for a particular book is 3/10,000 = 0.0003, or 0.03%. Very low! But of course, you don't submit to a single agent. Assuming no correlations between agents' response probabilities (which is probably true to a decent approximation), your chances of getting all rejections are (1-0.0003)^N, where N is the number of agents you submit to. Let's say you submit to 100 agents (definitely on the high end, but we're imagining that one is going all out on trying to get this published). Then the probability of getting all rejections is (1-0.0003)^100, which comes out to about 97% - in other words, there is a 3% chance of getting at least one offer of representation.

Estimates of how many agented books are accepted by a publisher seem to be all over the place. This article (https://jerichowriters.com/if-an-agent-accepts-your-work-what-are-chances-of-getting-published-2/) first claims that a good agent sells 2 out of every 3 books (67%) they take on submission. But then it considers things from the publisher's end, where only about 1% of books submitted get published, which means that if an agent submits to about 10 publishers, there is a (1-0.01)^10, or about 10% chance that a given agented book will be published. Let's split the difference and assume about a 35% chance that a book represented by an agent will find a publisher. That leaves us with a final probability of 0.03*0.35 = 0.0105, or just a tad over a 1% probability that a completed book will end up traditionally published. Not great, but not quite as improbable as winning the lottery!

I'm curious whether anyone else has tried calculating this number, or if there are places where my estimates seem off or could be improved.

I think your calcinations are not far off . Possibly I'm a bit more optimistic and say it is probably over 2% of submissions get published. I also believe a lot of writers have little or no knowledge of the publishing business . It is this that is the biggest reason most writers are rejected .

 

[Calla Lily]

<Mod Hat On>

Being 100% practical here, while speaking from the following place: 12 trad-pubbed books + many anthologized short stories. Looking for a new agent, so am in Query hell with a midlist track record. Book I'm querying is 180 degrees from everything else I've written. It takes courage and confidence to keep this up.

In my opinionated opinion, researching statistics is the way to make oneself give up. Here's my suggestion for channeling that energy:

Concentrate on the book you want to write. Revise it within an inch of its life. Send it to betas. Revised it some more. Write the query and synopsis and gird your loins for the QLH subforum. When it's as ready as it can be, start querying.

I treat writing as a business.

I know that isn't fun, shiny advice.

I recommend chocolate and the occasional massage.

</Mod Hat Off>

 

[Woollybear]

I think I was about 6 months into querying when I learned that agents can receive 4000 queries for each author they sign. It changed my expectations considerably, and put those expectations into a much healthier place. For this reason I think the numbers in the OP can be useful to think about.

Of those 4,000 queries, one of which, on average, leads to a signing, I disbelieve the other 3,999 manuscripts are unsellable. Even if 3900 are unsellable (which seems high), the agent is still only signing 1% of the good 100. (Even if 3990 are unsellable, the agent is only signing 10%.) So, these numbers shifted my thinking, hard, away from 'agents know what is publishable.' Add to that the various reports of occasional agents acting strangely, harmfully, illegally, and add, further, the high turnover in agenting, my personal view has shifted further yet, to 'agents sure are a funny mix, aren't they?'

I suppose writers are too. But, familiarity with the numbers and the accounts of bad actors has been good for my mental health. So, those numbers, in my opinion, are useful to think about, although the caveats that follow are totally legit.

 

[bunny hugger]

The more assumptions you make, the rougher the estimate is, but ts still a place to start. I am not diving into publishing a wide-appeal genre novel, so I take a bit of a different approach. For specialist non-fiction, I will start developing a book when I think there are at least even odds (50/50) that I can get it published somewhere reputable. That involves seeing a gap in the market, feeling qualified (alone or with co-authors) to fill it, and doing some reach outs of acquisitions staff with a one-page outline, or something along those lines.

 

[dickson]

ChaseJxyz's mention of the Drake equation reminded me of all those silly equations on the probability of life arising on Earth.

It's 1. The probability of life arising on Earth is 1.

Anyway, I return you to your normal programming.

That’s the posterior probability. The cat is already dead. Or alive. The quantity we ache to know in this thread is the prior probability, cos’ the penny is still in the air. I’d mix in a few more metaphors, but think the train has left the station.

 

[Aiwendil]

I must admit, I'm a little perplexed at the response this has gotten - and particularly the fact that people seem to be misconstruing my post as being defeatist or as claiming that publishing success is arbitrary and random. Expressing something as a probability does not mean that it is completely random! Of course the particular qualities of the book make a difference. Of course having a strong query letter improves those chances. If you have good reason to believe that your book and query are more publishable than the average, then of course your chances are better than the raw number.

I simply found myself wondering: what percentage of all books that are written and intended for traditional publication end up being published? This was just my thought process as I tried to answer that question. (Incidentally, I do reject the notion that it's somehow wrong to apply mathematical reasoning to a question like this. But look, I'm a physicist; this is just what I do.)

You can't assume that everyone goes through an agent. You can't assume that a debut author will have the same chance as a mid-lister or that either of them will be in the same boat as field leaders.

These are good points, but I expect these are small effects. Maybe I'm wrong, and if there are some numbers on these things, I'd be interested to see them - but my understanding is that the number of unagented books accepted by (mostly small) publishers is quite low, and that the vast majority of submissions an agent gets are from debut authors. I guess I also should have been clear that my intention was to consider debut novels - I realize now that I didn't explicitly say that.

There is a possibly larger correction here in that the estimates I was using for the number of submissions a publisher accepts presumably include a lot from non-debut authors, which actually means that the relevant number is lower than the estimates, if we assume that publishers are more likely to accept a submission from an established author than a new one. I expect that could be a large effect - I've done a quick search for numbers on what percentage of published books are debuts, but can't seem to find anything.

However, there is an undoubtedly significant effect in the other direction that I failed to notice:

You also can't assume that someone will keep querying if they get an acceptance.

Also keep in mind a portion of queries agents receive are for manuscripts that haven't even been written yet, because some people try to bounce ideas off agents.

This is a great point. I noted that assuming no correlation between agents' responses was probably a good approximation, but I hadn't considered the fact that this skews the numbers from the agents' side of things significantly. If we assume that a book that gets a positive response from one agent is more likely to get a positive response from another agent (a good assumpion), that means that since an author will typically stop querying after an offer of representation, the sample of submissions an agent gets is biased toward submissions that they will be more likely to reject.

To properly incorporate this effect would require a lot of information we don't have, but we can estimate the maximum size of it, if we make the (unrealistically generous) assumption that a book accepted by one agent would definitely be accepted by any other within the submission pool. Then if there are 100 agents in the submission pool for a given book, then if an agent accepts 3 queries per year, the maximum result of this effect would give 300 acceptances per year per agent if the queriers did not stop querying after an acceptance. That would mean an acceptance rate of 300/(10000+300), or about 3% instead of 0.03%. Again, obviously, this is an upper limit on the size of this effect and the real number is presumably significantly lower than 3%, but this does show that the effect could be large.

Again, though, please don't take this as anything but an intellectual exercise, and if you don't like applying math to things, please just ignore it!

 

[lizmonster]

(Incidentally, I do reject the notion that it's somehow wrong to apply mathematical reasoning to a question like this. But look, I'm a physicist; this is just what I do.)

As I'm the one who said this: it's not wrong. It's just irrelevant.

Publishing is 95% luck. Roughly. But that's once you reach a level of competence. Anecdotally - the vast majority of books queried are not at that level of competence (although it's admittedly a fuzzy line). The statistics are going to look much different if you could manage to sort above and below that line.

Also, my brother is a physicist, so I get it.

If we assume that a book that gets a positive response from one agent is more likely to get a positive response from another agent (a good assumpion), that means that since an author will typically stop querying after an offer of representation, the sample of submissions an agent gets is biased toward submissions that they will be more likely to reject.

I don't think this is a valid assumption. People don't generally query in a linear fashion - that is, they've got multiple queries out at once. And my ideal agent (on paper) isn't the same as someone else's - that is (once again) subjective. There's no way to scientifically predict which agents are going to get which subs.

 

[lorna_w]

I thought it was interesting, aiwendill. I like playing around with numbers.

I've been writing for a hellalong time now (first pro magazine story published 35 years ago) and I've been active online in writer's forums since there was such a thing, and before that was always in one or two critique groups. As a result, i've seen a lot of queries and first five pages, and have concluded most of them are, if not awful, just without any spark. A good query and well-written first five pages, by someone who has been at it a while, and probably had a handful of story publications in paying markets but no big press novel publications, is a fairly rare thing, in my experience. Even those good ones probably only have a 1 in 50 chance of getting a big five (four? where are we on that?) contract by the end of the query/agent process. That's only a guess + experience.

How many novels are Big 5 published every year in the US? That's a surprisingly hard number to find. Let's say 30,000 (though that may be high). Let's say 10% are by new authors. So, if my guesses are good, every good query is competing for 3,000 slots.

The numbers after that, which I've heard for ages and never saw refuted well, is of the 3,000 new authors, many of whom will get only a $5,000 advance, half will not earn out that small advance and probably won't be offered a second contract. So if you look at it that way, there are 1500 shots at a writing career every year, which I suspect is the goal of most writers (not just one and done).

You have to work hard and be good at your craft and business to be lucky, but there is luck involved in being one of those 1500. You control what you can, and accept you cannot control it all.

 

[Unimportant]

Novels are a weird thing in the 'art' world. Heaps of people do paint by numbers for fun, or do the occasional watercolour of a sunset to hang on their wall, or whatever. The vast majority of them never even think about having their work hanging in a gallery with a million dollar price tag. Heaps of people knit afghans and baby blankets, or play scales on the piano, or string bead necklaces, or whatever. They don't expect to become famous artists.

But pretty much every person who does NANO or types 50K words of story-ish stuff on their computer wants to be published by Random House, and they fling their books at agents and publishers willy-nilly. And the vast majority of those novels are as unpublishable as my ten-thumbs knitting and tone-deaf piano playing.

So, yeah, one in a thousand written novels gets picked up by a small press, and one in ten thousand gets published by a major house, and one in ten million becomes the Rowling or King bazillionaire best seller. But I only use those stats when some idiot says to me "Oh, you write? I was thinking about writing a book. I'd probably get a hundred thousand dollars for it, yes? Or would it be more?" And I explain the numbers to them so that they don't waste their time.