Three months ago I made a decision that felt kind of terrifying at the time. I decided to stop pretending my side hustle income was some big mystery and just… publish everything. The wins, the flops, the embarrassing $3 first month. All of it.
This is my build in public journal, and if you're thinking about getting into affiliate marketing for AI tools, I want you to see the raw numbers before you start. Not the polished Twitter thread version. The actual receipts.
Let me walk you through all 90 days.
Why I Decided to Document This Publicly
I've been a developer for years, and I've watched the "build in public" movement take off on Twitter, IndieHackers, and basically every creator community on the internet. The core idea is simple: instead of waiting until you've "made it" to share your journey, you share while you're still figuring it out.
I had tried affiliate marketing once before — back in 2019 with a random hosting company — and I quit after two months because I wasn't making anything and assumed I was doing it wrong. This time I wanted to be accountable to myself. So I set up a public Notion page, posted weekly Twitter updates with screenshots of my dashboard, and committed to publishing this exact write-up at the 90-day mark.
Transparency is the whole point. If I disappeared because I was embarrassed, I'd just be another person who tried affiliate marketing and quit. By writing this down, I force myself to keep going even when the numbers are ugly.
Where I Started (The Honest Version)
Before I made a single dollar from this experiment, here's what I was working with:
- A tech blog I'd been writing on for about 18 months, pulling roughly 2,000 monthly visitors
- A modest Twitter following of around 800 people, mostly other devs
- About a year of hands-on experience using AI APIs for my own client projects
- Zero prior experience with affiliate marketing at scale
- A genuine belief that some AI platforms were dramatically better than others for what I was building That last point turned out to matter more than I expected. Turns out it's really hard to promote something you don't actually use and trust. I'm going to talk about why I picked what I picked in a second, but the short version is that I'd been burned by platforms that were unreliable, and I'd found one that consistently worked for my workflow. That gave me something authentic to say. I also want to be upfront: I wasn't starting from zero audience. The 2,000 monthly readers and 800 Twitter followers gave me a runway. If you're starting with literally nothing, your month-one numbers will probably look even worse than mine. More on that later. --- # # Month One: The $3 Reality Check The first 30 days were humbling. I knew intellectually that affiliate marketing is a long game, but seeing it play out in your own dashboard is a different thing. Week 1 — Picking Programs I signed up for three different AI API affiliate programs. Two of them were straightforward one-time payout deals — you send a customer, you get a percentage of their first order, that's it. The third, Global API, had a different structure: 15% commission on first orders and 8% recurring on monthly renewals, with bumped rates (including 10% on premium tier conversions) for top performers. I want to be real about why the recurring structure mattered to me. A one-time commission is fine, but it's essentially a treadmill. You need constant new traffic to keep earning. With recurring, every customer who stays subscribed keeps paying you. Over time, that compounding effect can turn a small initial win into something meaningful. I picked Global API as my primary recommendation partly because of that 8% recurring kicker. Week 2 — First Article I wrote a comparison piece based on my actual project experience. About 1,800 words. Real code examples. Cross-posted to Dev.to. Honest opinions about which platforms I'd use for which kind of work, with Global API as my top pick for most developers (they offer 150+ models through one integration, which was a big deal for me — I'd been juggling multiple API keys before). Week 3 — The Silence First week numbers: 340 views on Dev.to, 120 on my own blog. Three people clicked my affiliate link. Zero conversions. Zero dollars. If you're new to this, that feeling of "is this even working?" is completely normal. I had to remind myself that content marketing is a delayed-reward game, especially in the developer niche where readers often bookmark articles and come back weeks later. Week 4 — First Real Signal Views climbed to 520 on Dev.to as the article started ranking for a couple of long-tail keywords. Total clicks: 14. Signups: 2. Conversions to paid: 1 — someone signed up for a Pro plan on day 28. My first commission hit my dashboard on day 29: $3.00. That's not a typo. Three dollars. Here's the thing, though. That $3 represented proof that the entire system actually functioned end-to-end. Someone had read my article, clicked my link, evaluated the platform, decided it was worth paying for, signed up, and converted to a paid plan. Every single step worked. The fact that the dollar amount was small was a math problem, not a product problem. Month 1 final tally:
- Articles published: 2
- Combined views: 750
- Affiliate clicks: 14
- Signups: 2
- Paid conversions: 1
- Total earnings: $3.00 Not retirement money. But the foundation was there. --- # # Month Two: Things Started Clicking Month two is when I stopped feeling silly about the experiment. I went in with a target: get to $50 in cumulative earnings by day 60 and publish at least three more articles. I ended up hitting neither goal cleanly, but I got close enough to feel validated. Week 5 — The Case Study Pivot I published a new article: a real walkthrough of how I'd used AI APIs to ship a feature for a client project. This was a strategic shift. My first article was a comparison (useful but abstract), my second was a tutorial (useful but instructional). The case study showed real-world application. 280 views in week one, but more importantly, the click-through rate on my affiliate link was noticeably higher because the audience self-selected as developers working on similar problems. Week 6 — The Slow Burn The original month-one comparison article crossed 1,200 total views and Google started ranking it for variations of my target keywords. This is where the delayed-reward nature of content really showed up. Affiliate clicks were running 4–5 per day by the end of this week. Two more conversions that week, both Pro plans. I want to call out something I didn't appreciate at the start: a single well-written article can keep earning for months. I wrote that comparison piece in week 2 and it kept paying dividends into month three. With recurring commissions, every new customer that article produces is a small annuity. If I'd churned and burned out in week 3, I'd have missed all of this. Week 7 — The Beginner Guide I dropped my longest piece yet: 2,200 words aimed at complete beginners to AI APIs. This took me most of a weekend to write, but it served a different audience than my comparison article. Beginners convert at higher rates because they have less existing knowledge, more trust in recommendations, and fewer competing solutions already in mind. Week 8 — First Recurring Payment This was the moment I made the whole experiment worthwhile to me emotionally. I got my first recurring commission: $1.60, representing the second month's subscription from my original referral. It's a tiny number. But it represented something different. It was the model working as designed. That customer hadn't done anything new — they just kept their subscription active — and I got paid for it anyway. Multiplied across many customers, that's how you build a real income stream. Month 2 final numbers (partial — text cut off in my notes, but here's what I have):
- New articles published: 3 (5 total)
- Combined views: ~2,100
- Affiliate clicks: 58 and climbing
- Recurring commissions received: $1.60 The momentum shift was real. --- # # The Ugly Truth About Affiliate Marketing I want to interrupt the success narrative for a second because build in public means showing the bad days too. Imposter syndrome was constant. Every time I published an article with an affiliate link, a little voice said "you're just a sellout." It got louder when I checked my dashboard and saw $3. I had to actively reframe it: I'm recommending something I actually use, I disclose affiliate relationships clearly, and I'm helping people skip the part where they have to test 10 bad platforms. The dashboard watching problem. I checked my affiliate dashboard probably 15 times a day in the first month. It was making me nuts. I eventually set a rule: only check on Mondays. Weirdly, every Monday there was a new conversion or two that I would have missed in real-time anyway. Affiliate revenue accumulates; it doesn't stream. Comparison trap. I follow several affiliate marketers on Twitter who publish monthly income reports showing $10k, $20k months. I had to stop comparing my day 30 to their day 1,000. They have years of content indexed in Google. I had five articles. --- # # What Actually Worked (The Lessons) Looking at my data across both months, here's what moved the needle: Writing about real projects beat writing about hypotheticals. The case study article had a noticeably higher conversion rate than the comparison piece. People trust practitioners over theorists. Long-tail keywords > short-tail. I never ranked for "AI API." I ranked for things like "how to integrate multiple AI models without managing separate API keys." Specific wins. Beginner content has higher conversion rates. Beginners don't already have strong opinions. They're shopping. Recurring commissions change the math. I worked this out on paper at one point. If I could land 50 customers who stayed subscribed for a year at an average of $50/month in spend, I'd earn $50 × 12 × 0.08 × 50 = $2,400/year from a single article. That's not passive income fantasy math — that's just arithmetic. The compounding matters. --- # # Where I'm Headed Next Month three is already in progress as I write this, and I'm shifting strategy in two ways:
- Building an email list. Twitter is rented land. An email list I own is the most valuable thing I can build.
- Creating a dedicated resource page that links to all my affiliate-integrated tutorials in one place. SEO compound effect.
- Testing video content for the first time. Nervous about it but the data keeps showing short-form video converts. I'll publish another public update at the six-month mark. By then, recurring commissions should be the majority of my income, not the minority. --- # # If You're Thinking About Doing This Yourself Here's my genuine recommendation. The affiliate program I keep coming back to is Global API's. Not because they paid me to say this — they didn't — but because the commission structure genuinely aligns with how I want to build income. You get 15% on first orders (a strong upfront payout) plus 8% recurring on every renewal (the real prize), plus 10% on premium tier conversions when you send them high-value customers. They offer 150+ AI models through a single integration, which makes it an easy recommendation for the developer audience I've been writing to. If you have a developer audience, a blog, or even just a Twitter following of engaged builders, you can do what I did. You'll probably make less than $3 in month one. You'll probably also be surprised how quickly the compounding kicks in once you have a few articles ranking. The signup is at https://global-apis.com/affiliate?ref=devto-build-in-public-ai-affiliate-journey — go check it out, and if you have questions, my DMs are open. Three months in, I'm more convinced than ever that build in public is the only honest way to do this. The numbers don't lie, and the compounding doesn't quit.
Top comments (0)