Everything forty-six books ever earned

$1,883.44 in fifteen years. And a best month I was about to tell you I couldn't explain.

This month my books earned seventy-four cents. One order. Thirty-three pages read.

That one's from the KDP dashboard. Everything else here comes from the payments report, which runs slower.

I could stop there. It's a good number, and I know exactly how it would read. But it's one month, taken in the last week of August. Building this issue on it would be the thing I published two weeks ago — quoting the figure that makes my point and leaving out the ones that don't.

So here's all of it.

The whole ledger

The earliest sales in my records are from January 2011. My first payment landed in December 2012.

Since then, across every book, every format, every marketplace, Amazon has paid me:

$1,883.44

That's the whole thing. Forty-six titles, fifteen years. It averages out at $40.94 a title, though I have no per-book breakdown and the real spread is certainly nothing like even.

The most recent twelve months of sales, July 2025 through June 2026: $391.03.

There are 102 months in which the books earned anything at all. Seventy-seven of them earned fifteen dollars or less.

One caution, because I'd get it wrong myself otherwise. A month with no payment isn't a month with no sales — royalties sit until they cross a threshold. I can tell you what I was paid and for which period. I can't tell you which months nobody bought anything.

What it was made of

Here's the part I didn't expect, and these books have been on sale since 2011.

Eighty-two percent of everything I ever earned came from paper. Physical books, printed one at a time, mailed to people. Audiobooks are another 2.3%.

The Kindle ebook business — the one the courses are about, the one with the funnels and the launch stacks and the also-boughts — is fourteen percent. Two hundred and sixty-three dollars, lifetime.

And Kindle Unlimited, where readers borrow instead of buying and you're paid by the page: thirty-eight dollars and seventy-seven cents, since my first page-read payment in September 2017.

I spent years treating ebooks as the business. Eighty-six percent of the money came from somewhere else.

Not because I chose badly. Because I never once opened this report and looked.

I nearly got this wrong here too, and it's worth showing you how. My first pass at these percentages said sixty and thirty-six. I'd taken Amazon's own column headings at face value, and "Sales" turns out to include everything sold through CreateSpace — which was print, Amazon's paperback service before it was folded into KDP in 2018. Four hundred dollars of paperback money sitting in a column I'd read as ebooks.

Then, in spring 2025

Look at the chart again. Years of near-flat, then one spike.

Three months over two hundred dollars.

Here's what makes that strange. The best month I'd ever had before that spring was $48.96, in October 2017. April 2025 beat my previous record by more than five times.

Then it decayed through the summer. Through 2026 the months have run between three and twenty-five dollars.

The draft of this issue said I had no idea what caused it.

I wrote that I hadn't launched anything. That there was no campaign in my records. That something had moved several marketplaces at once and I wasn't the one moving it. I had a whole section built on it — an unexplained success is worse than a flat line, a flat line tells you to stop, this told me nothing.

Then I had the numbers checked properly. Line by line, against the same file I'd just called my source.

What was actually in the file

April 2025 is the first month audiobooks appear anywhere in my records. Not a good month for audiobooks — the first month. Two audiobook royalty types both start there and had never existed before.

March 2025 is the first month UK page-read royalties appear. Also the first time, ever.

And Amazon.nl turns up in April 2025 with one ebook sale of $1.73 — the only appearance of that marketplace in fifteen years.

Three separate firsts, clustered on the two months either side of the best month of my life.

So: something changed in or around my catalogue that spring. I can't reconstruct what, from the royalty file alone. I don't have it in my notes either. What I can say is that the file was never silent — it had this in it the whole time I was drafting a section about how it contained no answer.

I was two days from publishing a mystery I hadn't earned.

Let me be exact about what I still can't claim.

I can't show that whatever changed caused the paperback spike. The audiobooks themselves earned about eight dollars in April. The Dutch sale was $1.73. The money was paperback — a hundred and ninety-three dollars of it, across the US, the UK and Germany. What I have is three firsts landing in the same two months as the spike. Correlation in time, and no mechanism I can demonstrate.

But "I have no idea what caused it" was false. And I'd have published it, in an issue about not overstating what your numbers support. There was a candidate sitting in the source and I hadn't looked hard enough to find it — because inexplicable was the more interesting story, and I'd already decided to tell it.

Same error as the table in Issue 04. Not a number I invented. A claim I hadn't earned, about my own data, pointing the way I wanted.

The advertising

Over the years I've put roughly a thousand dollars into advertising these books. That's my own estimate, not a reconstructed figure — I no longer have the reports, and I can't tell you which months the money went out. So I can't claim it produced the spring, and I can't rule it out either.

What I can put side by side: lifetime royalties $1,883, lifetime advertising about $1,000. On that estimate, roughly half of everything the books ever earned went back to the platform that sold them.

The arithmetic I should have done first is simple. A self-published ebook returns a royalty of a few dollars at most. Advertising is priced per click, and only a small fraction of clicks become purchases. So you need a lot of clicks, at real money each, to earn back a few dollars.

That isn't an optimisation problem. It's the size of the gap. You can improve targeting a great deal and still pay more for attention than the attention returns. I ran ads for a while and stopped. Stopping was right. I never wrote down why — which is how I nearly started again.

What this cannot tell you

Now the part that runs the other way, because it would be easy to write this as a tragedy and I don't think the data supports one.

One order this month is not evidence the books are bad.

It's not evidence they're good either.

At this volume there's almost no signal about quality. The numbers tell me very few people bought the books. They can't tell me whether the people who never found them would have liked them.

And sales aren't a clean measure of anything single. Price, cover, title, subject, positioning, reviews — they all sit between the writing and the number. I read that number as a verdict on my writing for years. It was never capable of delivering one.

One caution about the genre rather than anyone in it. The incentives all point one way. If low numbers mean your book is bad, the fix is craft, and craft can be sold to you in a weekend. If low numbers mean nobody has heard of you, the fix is distribution, and nobody can sell you that.

Why I'm not smiling

I read the same feed you do. Somebody's launch went well. Somebody's course changed their life. Somebody hit a list.

Most of it is probably true. I want to say that clearly, because it would be cheap to imply otherwise.

What's missing isn't honesty. It's the denominator.

You see the people a strategy worked for. You don't see the count of people who ran the same strategy and got $391 a year — not because anyone is lying, but because that post doesn't get written. There's no reason to write it and every reason not to.

So a picture assembles itself out of true statements and comes out false. I sat looking at it for years wondering what was wrong with me specifically.

Nothing was wrong with me specifically. I was comparing my whole ledger against other people's best months.

What survived

Open the report. I had this data for fifteen years and read it properly for the first time this week, because I needed it for a newsletter about honesty. Eighty-two percent paperback. Thirty-eight dollars from Kindle Unlimited. Three unexplained firsts sitting in the spike month. All of it available the entire time.

Publish the denominator. A result without the count of people who attempted the same thing is close to meaningless. When I don't have the denominator, I'll say so.

Writing more was never the lever. Issue 03 was about having no mechanism by which anyone would find the work. This is what that costs, in dollars.

The average was forty-one dollars a title. That doesn't mean the forty-seventh book would have earned forty-one — I've already said the spread is nothing like even. It means I had no evidence that writing another one would change the mechanism. And I'd have written it anyway, and called it a strategy.

What I'm doing instead is this. One name, one schedule, people who asked for it. In Issue 03 I said I'd tell you at Issue 26 whether it worked, with a number, whichever way it goes.

Now you know what the number has to beat.

What's the number you've never actually added up?