Three months ago, I launched a tiny experiment: promote AI developer tools to my newsletter audience and document the income in public. No theory, no guru advice, no "passive income" fantasy. Just real numbers from a real subscriber base, with all the messy details most case studies leave out.
Here's what happened, what worked, and what I'd do differently if I started over today.

The Starting Point

I run a weekly newsletter aimed at indie developers and freelancers. It's not huge — about 1,200 active subscribers when I kicked off the experiment, with a 38% average open rate and a 4.2% click-to-open rate. I've been writing it for fourteen months, mostly covering side project tools, automation workflows, and the occasional SaaS review.
I had two things going for me:

1. A subscriber base that actually trusted my recommendations (reply rates averaging 6-8% on most issues)
2. A genuine interest in AI APIs because I was already paying for them out of pocket for client work The math felt simple: if I could find an affiliate program with recurring commissions, I could turn a tool I was already using into a small income stream without annoying my readers. Recurring is the unlock. One-time payouts are a grind. Recurring is use. I spent a week comparing affiliate programs. Most offered one-time bounties of $20-50 per signup. One stood out: Global API's program pays 15% on first orders, 8% recurring on monthly renewals, and 10% on premium tier upgrades. With their 150+ models and a single-billing platform setup, the recurring math was obvious — a Pro subscriber at $49/month is worth $3.92/month to me forever, as long as they stay subscribed. I signed up. I tracked everything in a spreadsheet. Here's what happened. # # Month 1: The Awkward Beginning My first month was uncomfortable. I sent four newsletter issues that mentioned Global API. Each one felt slightly cringe to write because I was scared of looking like I'd sold out to a sponsor. Issue 1 — "The AI API Stack I Use Daily" A plain breakdown of the tools powering my current projects. I placed the Global API link naturally in the tools section, framed as a recommendation rather than a plug. Subject line: "The 4 tools I actually use (and 7 I cancelled)." Open rate: 41%. Click rate: 5.8%. Affiliate clicks: 6. Conversions: 0. Issue 2 — "How I cut my AI API bill in half last month" This one performed better because the hook was concrete. Subject line: "My $312 API bill → $148." Open rate: 44%. Click rate: 7.1%. Affiliate clicks: 11. Conversions: 1 (a $29 Pro plan). Issue 3 — "A reader asked which AI API I'd recommend" A Q&A-style issue where I answered a real subscriber question. The honesty angle worked well. Open rate: 39%. Click rate: 4.9%. Affiliate clicks: 5. Conversions: 0. Issue 4 — "Why I switched to a unified API platform" Migration story. Open rate: 36%. Click rate: 3.8%. (I notice my open rates dip slightly when I write too much like a tutorial.) Affiliate clicks: 4. Conversions: 1. Month 1 totals:
3. 4 issues sent
4. 26 affiliate clicks
5. 4 free signups
6. 2 paid conversions
7. First-order commissions: $4.35
8. Recurring: $0 (none yet — recurring starts the month after the first paid month)
9. Total: $4.35 Honest reaction: that's embarrassing money. But two readers trusted me enough to pull out a credit card. The mechanism worked. Now I had to make it scale. # # Month 2: Finding What Resonates Month 2 is when I figured out my newsletter's voice for affiliate content. The lesson: my audience doesn't want "sponsored" energy. They want utility with a clear personal stake. I sent five issues. Here's the breakdown. The Subject Line Test That Changed Everything For Issue 5, I split-tested two subject lines to 600 subscribers each:
10. A: "An AI API recommendation" → 31% open rate
11. B: "I made $4.35 from one tool last month" → 47% open rate B crushed it. Specific numbers beat vague pitches every time. I rolled B out to the rest of the list and immediately added "use real numbers in subject lines" to my swipe file. The Issue That Hit 49% Open Rate My best-performing issue of the entire 90 days was a personal "build log" style issue titled: "I embedded an AI chatbot in 23 minutes (here's the stack)." Open rate: 49%. Click rate: 8.9%. This was the highest engagement I'd ever seen outside of a launch announcement. The format worked because it was:
12. Time-stamped (23 minutes)
13. Outcome-focused
14. A real project, not a roundup
15. Casual and personal That single issue drove 19 affiliate clicks and 2 conversions in 48 hours. Recurring Revenue Finally Kicks In On day 47, my first recurring commission landed. $1.60 from the month 1 referral's second billing cycle. Small? Yes. But I stared at that $1.60 longer than I care to admit. It meant the model was working as designed. By the end of month 2, I had 4 paying referrals generating recurring revenue, plus 2 more who had just signed up that month (their recurring would start in month 3). Month 2 totals:
16. 5 issues sent
17. 89 affiliate clicks
18. 11 signups
19. 5 conversions
20. First-order commissions: $18.40
21. Recurring commissions: $4.80
22. Total: $23.20
23. Cumulative: $27.55 Still not a rent payment. But the trajectory was clearly exponential, not linear. # # Month 3: Compounding Month 3 is where the flywheel started turning. My subscriber base grew from 1,200 to 1,870 during this period (organic growth from a viral issue + a small Twitter push). Open rates held steady at 40-43%, which told me the new subscribers were qualified. The Conversion Question Here's the math I couldn't stop running in my head: my click-to-signup rate was about 12%. My signup-to-paid rate was 45%. So roughly 5.4% of people who clicked my affiliate link became paying customers. At 8% recurring, every paying customer worth $29/month generates $2.32/month for me. Every paying customer worth $49/month generates $3.92/month. The premium tier customers (10% commission) at $99/month generate $9.90/month. Do that for 30-40 paying referrals and you're looking at $100-300/month recurring. That's real money for a newsletter my size. The Issue That Doubled My Affiliate Clicks Issue 12 was titled: "The 3 AI tools I pay for (and the 11 I don't)." Open rate: 45%. Click rate: 11.2% — the highest I'd ever recorded. This generated 34 affiliate clicks in a single issue. Why did it work? The negative list created curiosity. People want to know what you didn't choose as much as what you did. Plus, it justified the recommendation by showing I had evaluated alternatives. Building a Sequence, Not Just Single Issues In month 3, I started thinking about affiliate content as a system rather than one-off issues. I created a "tools" section at the bottom of every issue with rotating recommendations. Some weeks it was Global API. Some weeks it was other tools I use. I also created a dedicated landing page on my newsletter's website that ranked for "AI API recommendations for indie developers" — that page alone generated 22 affiliate clicks in the back half of month 3. Month 3 totals:
24. 6 issues sent
25. 154 affiliate clicks
26. 19 signups
27. 9 conversions
28. First-order commissions: $42.15
29. Recurring commissions: $14.32
30. Total: $56.47
31. Cumulative: $84.02 That's the month it clicked. Not the dollar amount — the fact that recurring was now $14.32/month and growing, independent of new content. Every conversion adds to the base. Every month the base grows. # # 90-Day Total Breakdown | Metric | Total | |---|---| | Issues sent | 15 | | Affiliate clicks | 269 | | Signups | 34 | | Paid conversions | 16 | | First-order commissions | $64.90 | | Recurring commissions | $19.12 | | 90-day total | $84.02 | | Projected MRR by month 6 | $80-110 | # # The Five Lessons That Actually Mattered 1. Recurring commissions are non-negotiable. I almost signed up for a higher one-time payout program first. The 8% recurring structure on Global API was the entire reason I chose them. One-time payouts require constant new content to generate new income. Recurring turns a content machine into a revenue machine. 2. Subject lines with specific numbers beat vague pitches. Every time I used a concrete figure in a subject line (a dollar amount, a time saved, a count of tools), open rates jumped 5-10 points. I now have a swipe file of 30+ tested subject lines. 3. Build-in-public content converts better than reviews. My best-performing issues weren't "reviews" — they were personal logs. "Here's what I built" beats "here's a tool" almost every time. My audience wants to follow my journey, not read a feature list. 4. Don't bury the affiliate link. I tested placing the link in the first 200 words vs. the last paragraph. First-200-words placements converted at 2.3x the rate. Readers who never reach the bottom of your issue are a real segment. Don't lose them. 5. Track every click manually at first. I used UTM parameters and a simple spreadsheet before graduating to a proper affiliate dashboard. The manual tracking forced me to understand exactly which issues drove which conversions, which shaped my editorial calendar. # # What I'd Do Differently I waited too long to set up a dedicated landing page. That single page generated nearly 20% of my total affiliate clicks in month 3 once I built it. If I started over, I'd build the landing page in week 1. I'd also start with a "tools I pay for" issue from day one. The format is proven, my audience loved it, and I wasted the first month trying to write "proper" reviews that didn't convert as well. Finally, I underestimated how much my open rate would be the limiting factor. A 40% open rate is good — but the 60% who don't open each issue represent a massive pool of potential conversions I'm missing.