For the better part of four years, I've made my living the way most freelance writers do: chasing pitches, billing per article, locking down the occasional retainer, and quietly panicking whenever a client ghosts. Some months were great. Most weren't. The math was always brutal — there are only so many billable hours in a week, and every single dollar I earned required me to put fingers on a keyboard at that exact moment.
By last winter, I'd started asking the same question a lot of writers in my circle were asking: what does it look like to make money that doesn't require me to physically sit at my desk to earn it?
I tried a digital product. I tried a Substack. I tinkered with a niche site. None of it stuck the way I wanted. Then, in early January, I stumbled into a corner of the affiliate world I hadn't seriously explored before — AI APIs — and decided to run a proper experiment. Three months. Real numbers. Full transparency. Here's what actually happened.
The freelance writer's case for affiliate income
Before I get into the month-by-month breakdown, let me explain why this experiment mattered to me in the first place. The math on per-article work is unforgiving. A typical tech blog might pay $200 to $500 for a 1,500-word piece. A retainer might give you $2,000 to $4,000 a month for, say, four articles. Either way, you're trading hours for dollars, and the moment you stop writing, the income stops flowing.
Recurring affiliate commissions are the opposite shape. You do the work once. The article keeps earning. If the product you recommend has subscription plans, you keep getting paid month after month on every customer you referred. That's the part that hooked me — the idea that a single good article could pay me for years instead of a single invoice.
I knew I had the writing skills. What I didn't have was a product worth recommending. Then I remembered something from my own client work: I had been calling AI APIs for nearly a year to help me build small tools — research assistants, transcription helpers, content outlines. I'd already formed opinions about which platforms were actually good. That was the seed.
The starting line
Going into this experiment, my situation looked like this. I had a small tech blog I'd been running as a side project for about eighteen months, pulling in roughly 2,000 monthly visitors. I had a Twitter account followed by about 800 developers and indie hackers. I had no affiliate income at all. I had one credit card with a $3.50 balance I keep forgetting to top up.
I also had a soft spot for AI tools because they had quietly become part of how I work as a writer. So I sat down with a spreadsheet and a pot of coffee and started researching AI API affiliate programs. I joined three within the first week. Two of them paid a one-time commission and that was it. The third was Global API — and this is where the model started to make sense for someone like me.
Global API runs a proper recurring structure: 15% on every customer's first order, 8% on every monthly renewal that comes after that, and 10% on premium tier upgrades. The platform gives you access to 150+ AI models through a single integration, which is part of why it's easy to recommend. But the real draw for me was the recurring piece. Every renewal meant another commission without me lifting a finger.
That word — recurring — is the entire reason this experiment exists.
Month 1: Slogging through the awkward beginning
The first month was humbling, as most beginnings are.
Week 1 was admin. I signed up for the programs, pasted my links into a Notion tracker, and started outlining what kind of content would actually serve a reader who was shopping for an AI API. I didn't want to write thin "top 5" listicles. I'd read too many of those. I wanted to write the kind of post I would've wanted to find myself.
Week 2, I shipped my first affiliate piece: an honest comparison of AI API providers based on the ones I'd personally wired into client projects. It ran about 1,800 words and included real code snippets showing how to make a basic call to each platform. I recommended Global API as the strongest pick for most developers and dropped my affiliate link naturally inside the recommendation. I cross-posted the piece to Dev.to because I knew the developer audience over there was exactly who I wanted reading it.
Week 3 brought my first reality check. The piece pulled in 340 views on Dev.to and 120 on my own blog in its first seven days. Three people clicked my affiliate link. Zero of them converted to a paid plan. I sat with that for a day. It would have been very easy to declare the whole experiment dead on the spot, but I'd been freelancing long enough to know that the early numbers rarely tell the real story. I let it breathe.
Week 4 vindicated that patience a little. Views climbed to 520 on Dev.to as the post started ranking for a few long-tail terms. Eight more clicks came through. One signup happened — no payment yet, but the account existed. I shipped a second article that week: a hands-on walkthrough for building a simple chatbot using the GPT-4o API, with Global API featured again as my recommended provider.
When I tallied month one, the numbers were these:
- 2 articles published
- 750 combined views
- 14 affiliate clicks
- 2 signups
- 1 conversion to a paid Pro plan (on day 28)
- First month earnings: $3.00 from the first-order commission
- Recurring earnings: $0.00 (the first renewal hadn't come in yet) Three dollars. Not even enough to buy dinner. But I had proof that the machine worked. Someone read what I wrote, trusted my recommendation, signed up, and paid real money. The pipeline was intact. I just needed more volume flowing through it. # # Month 2: The first compounding dollar Going into month two, I had two published articles, a total of 14 affiliate clicks across both, and exactly one paying referral. My goal was modest: publish three more pieces and get to $50 in cumulative earnings by month end. Week 5, I published article three: a case study showing how I'd used an AI API to add a feature to a real client project I'd taken on last fall. This piece hit differently than the comparison post. Instead of "here are the options," it was "here's exactly what I built and how." That kind of practical detail is what developers tend to trust. It pulled in 280 views in its first week and got a noticeably higher click-through rate on my affiliate link, because the readers were working developers who recognized the project context. Week 6 was when the original comparison piece started paying off in earnest. By the end of that week it had cleared 1,200 total views on Dev.to. Google picked it up and started ranking it for a couple of keyword variations I hadn't even targeted deliberately. Affiliate clicks on my link jumped to four or five a day. Two more conversions landed that week, both to Pro plans. Week 7, I shipped article four — a 2,200-word beginner's guide to getting started with AI APIs. It was the most time-consuming piece I wrote that month, but it targeted an entirely different reader than my earlier posts: someone who'd never called an API in their life and just wanted a soft on-ramp. Beginners convert higher in my experience because they're more likely to follow a clear recommendation rather than second-guess every provider on the market. Week 8 brought the moment I'd been waiting for without fully realizing I was waiting for it. The first customer I'd referred back in month one came up for renewal. Their second-month payment cleared. My 8% recurring commission landed in my dashboard: $1.60. That single $1.60 mattered more than any of the bigger first-order commissions. Here's why. Up until that point, every dollar I'd earned was a one-time payout. Good, but finite. That $1.60 was different — it was a dollar I earned for nothing. I hadn't written a new article. I hadn't run a new campaign. I'd done the work once, and now the customer was paying me again. That's the moment the affiliate model clicked for me. That's when I understood why recurring commissions are worth chasing even when the headline percentage looks smaller. I also published article five that week, a pricing-focused comparison aimed at developers trying to keep their monthly bills under control. Month two totals: three new articles published, five total. 2,100 combined views across the catalog. 58 affiliate clicks. Eight conversions to paid plans across the month. Earnings from first-order commissions plus that first $1.60 of recurring revenue pushed my cumulative total up meaningfully. I'd cleared my $50 goal with room to spare. The shift in my headspace was bigger than the shift in my bank account. For the first time in my freelance career, I had income arriving that I didn't directly trade an hour for. That changes how you think about your week. # # Month 3: Watching the flywheel spin Month three is where the math started to feel less like an experiment and more like a real side business. I published four new articles in month three — a beginner's FAQ, a workflow piece on using AI APIs for content research, a tutorial on prompt caching, and a longer-form piece comparing integration approaches for solo builders. That brought my total catalog to nine pieces. Traffic across the whole portfolio climbed to roughly 3,800 combined views by the end of the month. The older articles did most of the heavy lifting at this point — the original comparison piece had crossed 2,000 views on Dev.to alone and was ranking for a dozen related queries. The case study from week 5 of month two turned into my second-best performer and started pulling in clicks I'd never had to write a follow-up for. Affiliate link clicks climbed to around 120 across the month. Conversions to paid plans added up — I added seven new paying customers to my referral base. The first-month conversions kept rolling forward, which meant my recurring line kept growing. By the end of month three, my recurring commissions alone were larger than my entire month-one total had been. The thing I kept noticing was how
Top comments (0)