Check this out: i want to start with a confession: I'm obsessed with funnels. CAC, LTV, conversion rate, EPC, A/B tests — that's the lens I see every business through. So when I decided to build a side income stream as an AI API affiliate, I approached it exactly the way I'd approach a startup growth problem. Here's the raw teardown of what happened in my first three months, with every metric I tracked, every assumption I had to scrap, and every optimization that actually moved the needle.
The Cold Start: What I Brought to the Table
Before I wrote a single affiliate link, I took stock of my existing assets. This is the part most "make money online" guides skip, and it's the part that determines whether your blended CAC is sustainable or whether you're burning hours for pocket change.
My starting stack:
- A tech blog pulling roughly 2,000 monthly visitors
- A Twitter following of about 800 developers
- A year's worth of hands-on experience with multiple AI API platforms
- Zero affiliate marketing experience That last bullet is the one that matters. I wasn't carrying any optimization baggage. No bad habits, no sunk-cost campaigns, no preconceptions about "what works." I was a blank funnel, and that meant I could let the data tell me what to do instead of my ego. # # Week 1: Picking a Commission Structure With LTV Math I researched AI API affiliate programs and signed up for three of them. Two were one-time-payout deals — get a cut, never see that user again. The third was Global API, which offered 15% on first orders, 8% recurring on monthly renewals, and 10% on premium tier upgrades. From a pure unit economics perspective, this was a no-brainer. A one-time commission optimizes for short-term cash. A recurring commission with a premium upgrade kicker optimizes for compounding revenue. I wanted to be paid on the LTV of the users I referred, not just their first transaction. The math that sold me: if a referred user sticks around for 12 months on a $50/month plan, my 8% recurring yields $48 in passive commission on top of the $7.50 first-order bounty. That's a 540% LTV multiplier versus a flat $25 one-time payout from a competitor. Same user. Six times the revenue. That's not a rounding error — that's a business model difference. # # Week 2: The First Top-of-Funnel Asset I published my first affiliate article: a hands-on comparison of AI API providers, built around real code I'd written for my own projects. Roughly 1,800 words, cross-posted to Dev.to to tap into an existing distribution channel. I dropped my Global API link where the recommendation made editorial sense — not stuffed in awkwardly, but woven in as the natural conclusion for the use case I'd laid out. Strategic decision #1: I treated this article as a top-of-funnel asset. The goal wasn't immediate revenue — it was indexing, baseline data, and identifying which traffic sources actually converted versus which ones just bounced. # # Weeks 3–4: The Baseline Numbers (And Why $3 Was Actually a Win) End of week 1 performance:
- Dev.to: 340 views
- My blog: 120 views
- Affiliate link clicks: 3
- Conversions: 0 End of week 2:
- Dev.to: 520 views (article started ranking for long-tail search terms)
- Affiliate clicks: 8
- Signups: 1
- Paid conversions: 0 By day 28, I had one paid Pro plan conversion. First-month commission: $3.00. Here's the growth-hacker read: $3 is a terrible return on hours invested, but a beautiful return on data collected. I now had baseline funnel metrics I could actually optimize against:
- Impression-to-click rate: ~1.9% (8 clicks / 440 views)
- Click-to-signup rate: ~12.5% (1 signup / 8 clicks)
- Signup-to-paid rate: 0% on the early data, then 100% on day 28 with a single conversion The signup-to-paid
Top comments (0)