Why not start an indie tech book publisher?

I’ve spent a fair amount of time over the last year thinking about the economics of book publishing now that I have sales trajectories on two book and have gotten to see a bit of the work behind Alex Xu’s System Design Interview – An insider’s guide, swyx’s The Coding Career Handbook, and Gergely Orosz’s Building Mobile Apps at Scale and The Tech Resume Inside Out. By now, I’ve spent enough time thinking about it that I wanted to run through the numbers once so that I can stop thinking about it.

So that’s this post’s central question: what would the economics of a small tech publisher look like? I write this with the hazy memory of my sister briefly starting and running a publishing company, which sounded much harder than any spreadsheet might imply.

Starting with the simplest imaginable model, the publisher makes about $31k USD in profit per book. If we assume the publisher only employes one person and that they’re living in California with an effective tax rate of 40%, then you’re looking at about $19k post-tax per book.

Let’s dig into these assumptions a bit.

AvgSales of 10,000 per book. There are a lot of numbers out there about typical technical book sales, for example, in 2007 John Resig wrote that, “The vast majority of technical books will never sell more than 4000 copies.” Conversely, of the books I’ve published and the other folks I’ve chatted with directly, 10,000 feels low, but there’s a lot of bias in my sample. If you think about the number of book proposals that ever become a published book, I imagine the ratio is much closer to 2% than to 20%. A publisher could invest a considerable amount of resources on books that never actually become a book.

Even ignoring the risk of unfinished books, there are a surprising number of good books out there that don’t get purchased in volume. As a few examples, I thought Fool’s Gold? was a great, surprisingly detailed book on the data behind angel investing (caveat: in 2008), but it only has 10 Amazon ratings. Hillel Wayne’s Practical TLA+ is an amazing book on a topic that deserves more attention, but has similarly few ratings. Amazon ratings are not, of course, an actual measure of sales, but I think they’re a somewhat useful proxy. My guess is that neither of these books sold 10,000 copies despite being objectively excellent books. This suggests a publisher could consistently select great books and still fail to meet this model’s criteria.

AvgPrice of $25. Folks buying technical books are relatively price insensitive, although effective price points do vary a bit by market. $25 for US market feels like a safe assumption, and you could probably push it a bit higher. If you wanted to spend more time here, the two interesting trends that I see is the gradual shift upward of book prices by publishers like O’Reilly and indie publisher who are driving up order size by attaching higher margin offerings to their books, e.g. join a book-driven community for $10/month.

% Expenses of 50%. The model is a bit overly simplistic here. A more useful model would split costs between fixed costs (editing, cover creation, etc) and variable costs (printing costs for physical book or platform fees for digital books). Your single biggest driver here is shifting folks off Amazon and onto lower platform cost services like SendOwl or Gumroad, which could easily shift platform costs from 70% of sales price down to 5%. However, this is a complex conversation, and even Holloway doesn’t sell exclusively through their own platform (e.g. Holloway’s Angel Investing: Start To Finish on Amazon).

Author Share (of profits) of 50%. I think this is a pretty plausible estimate. There is a very wide range here from Leanpub’s 80% author share to the significantly less that folks seem to get from larger publishers ($3-5k advance that they may not earn out from). Part of the wide range is because publishers do or don’t provide a wide range of services: cover creation, printing, editing, and so on. Leanpub is cheap, but provides few of those services. O’Reilly is expensive, but provides all of them.

Altogether, I think that these are pretty reasonable estimates. Extrapolating from my experience watching Stripe Press’ creation, I think that it’s fairly reasonable to do four books a year with a team of two, although it would require a fair amount of outside contracting for editing or whatnot. Let’s extend the model slightly to better capture those fixed costs.

Adding the fixed costs doesn’t change the outcome too much, although it certainly does if you drop the average sales number down from 10,000. That being said, if we want to pay our two employees a salary of $75k/year, we need to publish 5.7 books per year and meet all the other assumptions embedded in the model (10,000 average sales, etc). If we had an average sales of 5,000 copies per book, then all of the sudden we need our staff of two to publish 14 books per year: brutal.

Altogether this captures why book publishing is so hard: it’s a bit hit driven, and as a publisher you need to capture much of the hits’ success to offset the costs of the books that don’t break out. There’s poor scalability because there are a bounded number of “good” books coming at any point, and marketing doesn’t have good economies of scale across a huge number of different books. I can’t think of how I’d change the model to make this look like a venture capital investable business, but the necessary assumptions would be hard to defend.

This brings us to the final top-of-mind topic: how does the “creator economy” tie into this? One of the great entrepreneurial traditions is selling shovels to folks pursuing a gold rush, and certainly there are more folks trying to directly own their own distribution, marketing, and content on platforms like Substack, and those platforms have been able to attract venture investment.

It seems likely to me that you could design a book publisher than positions itself to benefit from the creator economy, but it likely wouldn’t look anything like the sort of independent publisher that I was initially thinking of. Rather, if we inspect the Substack model a bit, I think a successful publisher here would look a bit like an iteration on Leanpub. For example, you offer a self-service self-publishing platform that offers a high author share, perhaps 80%. You add tools for authors to write and market their books. Using the analytics from those tools, you can determine which books are likely to perform highly, and then approach those authors individually to offer a different contract where you onboard them into a hands on process with a lower author share (perhaps 40%) and significantly more services (cover creation, editing, etc).

Following that model, you could derisk all the various risks discussed above (books that don’t finish, books without an audience, books with authors who don’t engage on marketing, etc), and have a high enough hit rate to make the model work to some extent. Even if you did all that, however, it honestly still doesn’t seem like a particularly great business.

That being said, I’ll readily admit that I personally would be very excited to work with a company that wanted to sign this sort of contract with me:

• I create, edit, create cover for, and market my book
• Sell my book on Amazon and elsewhere in physical and digital versions
• Sell my book internationally and deal with licensing rights
• Collect errata and feedback
• If my book is unavailable for sale in print or digital version for more than four months, rights return to me to work with another publisher
• Send me a 1099 and payment once a year
• I’d personally be glad to split 50%/50% on profits with a company doing this

What’s less obvious to me is if there are enough folks out there who want this sort of setup. I really just want to write books, I don’t want to deal with the distribution details. I’m even glad to give up a reasonably large chunk of revenue to avoid those distribution details. Conversely, I also want to do a bunch of specific things that specific publishers disagree with, like have almost the entire book available online for free.

Supporting this sort of business is operationally intense but doesn’t really require any technology platform to speak of (e.g. I did all of those things myself for Staff Engineer when I self-published it, and although it was confusing I did get through it in the end). That said, once again it’s not an investable business, even though it might generate a fair amount of revenue with a healthy margin so-long as you were able to acquire good authors.

Anyway, ending this thought exploration here. It doesn’t look like I’ll be starting an independent publishing business anytime soon (sort of my version of the doomed trope of opening a cafe “for fun”).