Self-publishing survey: what it does and doesn’t tell us

Author Beverley Kendall set out to dispel the myth that only a handful of authors were doing well self-publishing – those making millions of dollars a year or those being offered substantial deals with traditional publishing houses. She ended up with a survey of 822 self-publishing authors and a ton of data, which she has compiled into a fascinating report which you can now read on her blog. If you’re an author who self-publishes or is thinking of self-publishing it’s well worth your while reading it. If you’re interested to know why so many authors are starting to self-publish some or all of their books, you’ll get a good understanding of that, too.

One of the headline stats is that over 200 authors who took part in the survey earn enough from their self-publishing to have quit their day jobs. That’s pretty awesome and it certainly proves her initial point. The range of income is also interesting, because it’s not all or nothing. There are authors in every band between $0-$10,000 and $500,000+. Some people are making a reasonable living, some are making a substantial supplementary income, some are pulling in a lot more than they could ever have dreamed of. Nearly half the respondents earned over $10,000 from self-publishing in 2013. If you simply want to be able to say, ‘There are hundreds of self-published authors earning very good amounts of money’, this survey has given you the evidence you need.

However, before we get too carried away and try to extrapolate from the survey to making more general statements, I do want to point out a couple of cautions.

First: this is a self-selecting sample.

One of the most important foundations of statistical analysis is the random sample. You can get surprisingly accurate results from relatively small samples, provided those samples are taken randomly. By contrast, if there is a bias in your sampling method, then it doesn’t matter how large your sample is, it will not be representative. 822 authors responded to Kendall’s survey, which is a good number. But there is absolutely no way of knowing how representative they are of the whole group of self-publishers. There may be geographical bias (are US authors over-represented, for instance?), language bias (how many, if any, respondents were publishing in a language other than English), digital bias (there are still some people doing print self-publishing). Almost certainly, there is survivorship bias.*

There is one form of bias that we can see from the survey results themselves, which is genre bias [p6]. I did wonder whether it would be possible to use the data from this question to weight some of the rest of the analysis. That could be useful with the genres which are best represented, but for the smallest genres, the actual numbers are too tiny to make it a significant sample. For example, 1% of the respondents write children’s fiction. That’s about eight authors. I don’t think you can conclude much of anything about self-publishing children’s fiction on the basis of this survey. Anyway, Kendall rightly points out that there is a large representation of romance authors, but she adds, “Having said that, I also believe that the percentages represented in the pie chart below is a reflection of what is selling, especially when it comes to digital books.” I have no way of knowing if that’s true or not, except in the very broadest terms, that romance sells more than any other genres. Maybe that’s all she meant to imply.

[Side note: I actually wonder whether there could be some useful data to be mined by extracting the results for the largest genres represented and treating that separately. What if the romance results were considered, for instance? Or even the different romance subgenres? I do realise that was well beyond Kendall’s goals and it’s not meant as a criticism of her report, just something I’d be curious about.]

Given the nature of the sample and its potential biases, I do not think it is possible to draw any statistical conclusions about self-publishing as a whole on the basis of this survey. I do think it is useful for identifying several other things, in a non-quantitative manner. But it cannot be used (and it should be said that Kendall does not do this) to calculate average self-publishing income or likely self-publishing income. It is not, sadly, proof that the oft-quoted, never-proven saying that the average self-published author makes $500 is not true. We don’t know, and for a lot of very good reasons, we probably never will (see below for why this doesn’t actually matter all that much).

Second: the survey does not compare self-publishing with traditional publishing.
Or at least not directly. What it does do, a little bit, is compare the earnings from various kinds of publishing (traditional, digital-first, self) for hybrid authors. It does NOT compare the income of solely trad-published authors with digital-first authors or with self-published authors. That is a different thing. For hybrid authors, self-publishing can have an effect on income from non-self-published books, just as their traditionally published books can have an impact on their sales of self-published books. You can see this in figure 3, where the respondents are divided according to their entry point into publishing and it seems clear that hybrid authors who began with traditional publishing do, in general, have higher incomes from their self-publishing than those who started with either digital-first or self-publishing. I actually think this insight into the value of being a hybrid author is fascinating.

But there is no data in the survey about solely trad-published authors and therefore no conclusions can be drawn about trad vs. self-publishing. To be clear, this is NOT a criticism of the survey. It wasn’t intended to do that and if that’s what you’re interested in, you’ll have to do your own survey.

What can we learn from the survey: the habits of high-income authors
So, if we can’t draw quantitative statistical conclusions from the data, and we can’t use it to compare self and trad-publishing earnings, what can we say? I think what we can do is talk about the ‘habits of high-income authors’. More precisely, the habits of the high-income authors who participated in this survey. These are more likely to be true of high-income romance authors than, say, high-income children’s authors because of the genre bias in the sample. They may only be true of the group of ‘high-income authors who have 3 degrees of separation from Beverley Kendall’ or ‘high-income authors who have time to procrastinate by filling in online surveys about their income’ (okay, that’s all authors). The reason I’m calling these habits is because it’s very difficult to be certain about causality – for instance, does spending money on a professional cover generate sales, or is it that higher earners have more spare money to spend on professional covers? The data from this survey doesn’t help us answer that.

The habits of the high-income authors in the survey are:
1. Using professional covers and professional editing
2. Writing books in a series
3. Making a book in a series free
4. Having a sizeable backlist
5. Pricing at $3.99 or $4.99
6. Having self-published for 4-6 years
7. Being traditionally published first

These things don’t necessarily cause authors to earn more (and not all of them are possible to control), but the authors who earn more tend to do those things.

Kendall has done an amazing job in collecting and analysing a lot of data and I think a lot of authors will be as grateful to her for that as I am. There’s much more detail in her report and I encourage you to read it carefully and work out for yourself what you can learn and apply to your own publishing career. And if you are a reader, then do take her reader survey here.

Why doesn’t average author income matter?

Because there isn’t an average author.

If you were thinking of going into, say, teaching, you could look at the average teacher’s income and it would give you a good idea of what you were likely to earn. That’s because teachers are all paid much the same amount. You’d know that if you were promoted to headteacher, or perhaps if you moved to a private school, or whatever, you’d earn more. But there’s a fairly limited range of salaries within which the vast majority of teacher’s incomes fall.

That is not how it works with writing. Some people are J.K.Rowling and some are me. Quite a lot earn even less than me. The range is huge and unpredictable. Over half of Kendall’s respondents earned less than $10,000 last year. But at the same time, a quarter earned enough to give up their day job. It doesn’t matter what the average author earns because, unlike teaching, that is not a reliable guide to what you are likely to earn. There are authors who only ever want to self-publish the one book of their heart and are happy to sell a dozen copies to their family and friends. Is their income relevant to you, if you are serious about writing as a career? There are authors who publish obscure guides to niche non-fiction subjects. Is their income relevant to you, if you are writing fiction in a popular genre? What might be more relevant is what average authors like you are earning – in your genre, with a similar rate of productivity, for a start. But even then, it’s hard to predict why one author sells a million copies of a book that seems much like any other. The average income is not a likely income.

*Several people on twitter pointed out, quite rightly, that all surveys of author income are self-selecting and that it would be virtually impossible to do anything else. I think that’s probably true. If Amazon ever released its kdp figures, that would be amazing, but until then this is probably the best we can do. I still think it’s important to understand the limitations of a self-selecting survey and to be careful about the conclusions which can be drawn from it.