In March 2026 I bought a section in Wānaka and put house plans in with the council. The mortgage is large. I've decided to try to pay it off with affiliate links, and I'm letting you watch. That's the stunt, and the real thing sits underneath it.
The question. Is the digital marketing playbook still true? In 2026, can one marketer build a network of sites written by a machine from his own seed, pay for it with affiliate links and one book, and have it work in the open, with his name on every page and the whole mechanism disclosed?
Why now. Years ago I recorded a course on typesetting books, but only after I'd stopped designing them. It was accurate and I was bored stiff, because I had already left; I was finished, not asking questions, and when the emails with questions arrived I didn't want to think about book design, so I shut it down. I'm not making that mistake again. I'm still a working marketer, in the accounts every week with other people's money, writing this while I'm in the trade and asking the questions as I go, and that is what keeps it alive. I'm also a digital marketer approaching forty in 2026 and asking what else I could do, and this is one way of putting a cap on that part of the career while I still have the standing to test it.
The ladder. My silly goal is to sell a million books. My less silly goal is to make a million dollars from books. My less-less silly goal is a million from books and a back end of software I've tested. All three are true at once, and the count today, October 2026, is zero, because the second edition has just launched and there is no date on the million. It is a goal held lightly, not a forecast.
The benchmark. I used to have affiliate programs as a very minor side hustle. At its best it made about $4,000 a year, and it died of admin, not tactics: I failed to keep up with the paperwork and the money dried up. If a machine can run the admin as well as write the pages, that is a finding on its own, and probably the more useful one.
Traffic and offer, applied to me. If I'd gone viral three years ago I'd have made nothing, because I had nothing to sell. A little before that I'd have made nothing either, because the only thing I had to sell was my own time, and time runs out. The experiment only makes sense now that the offer exists: a book I'm finally proud of, and a back end of software I've tested. Traffic without an offer is noise you paid for.
The constraints. A low-value single book does not carry the online ecosystem on its own, so I've worked within my constraints: no manufacturing, no stock, no third-party logistics, no sprawl of products, nothing that wakes me at night. Affiliate links, my own book, and a machine that writes and administers all pass that test.
The rules. Everything is generated, not edited. If I don't like something, it gets regenerated, not fixed. The seed is mine: the thinking, the frameworks, the one true thing I'll say about each tool. The point is generated content, not generated-then-edited content pretending to be human, so I don't promise that a person reads every page on every regeneration; the rules do, and the seed carries the promise. I don't post slop at volume, and what I write under my own name on LinkedIn and Substack is written by me; nothing from this machine goes out there as mine.
What I'm betting. My own Trust Algorithm says a network that vouches for itself is brand dressed up as reputation, and that faked trust works for a season and then collapses, so I'm not predicting this lasts forever. I'm making three smaller bets.
- You can make good money in a dying market. Print was shrinking when BookPrint turned over a little under a hundred thousand dollars in the year to April 2013 and nearly double that the year after. There's a window, and timing the entry is the skill.
- Does it still work if it's fully transparent? One named author across every site, every link that pays marked as paying, and this page explaining exactly what the network is. That's not a hidden network, it's a publisher who runs several sites, and they link each other openly because one person wrote them. Nobody knows which way the algorithms read that, and that is the experiment.
- Does the loop appeal to AI? Five sources that agree look like consensus even when one person wrote all five. It probably helps, for now, the way these things always help until the systems learn to weigh independent sources, and I'd rather be the one who said so than the one it catches out.
What I watch, and when it ends. Income and views on the money side; good conversations and deeper thinking on the personal side, which on their own would keep it running while the time cost stays small. There is no end date. It ends if marketing fundamentally changes or the websites go to zero, and not before, because on corporate or slower work you can't test new tools and ideas against a client's brand equity, and your own experiments have always been part of this trade, especially in search.
If the search traffic collapses, the experiment doesn't. The content is owned, the book exists either way, the writing doesn't depend on Google, and AI answers are their own channel. One leg can fall off and I'll write that up too.
This is a sense-making exercise pointed at myself. The whole universe is about seeing the system you're standing inside, and this is me standing inside mine, on purpose, with the lights on.
Changelog
- 11 October 2026: the page is written. Three bets open, no end date, the book count at zero. The AI-visibility baseline is to be recorded on the day this goes live, before anything else changes.