I track everything in a Notion table. Every dollar, every hour, every link click. Some people call it obsessive. I call it the only way to know if a side hustle is actually worth my time. Because here's the thing — if you're grinding away on nights and weekends while holding down a day job, you need to know the real hourly rate. Not the hype number. The real one.
That's why I started logging every minute of my AI API affiliate experiment. Three months in, I've got enough data to actually break it down. Let me show you the math.

The Setup: What I Already Had (And Didn't)

Before I dropped a single affiliate link, I needed to be honest about my starting position. I'm a backend developer pulling a decent salary at my day job, but I've always had that itch for side income that doesn't feel like a second shift at McDonald's. You know the feeling.
What I had going in:

• A tech blog I'd been writing on for about 18 months, getting roughly 2,000 monthly visitors
• A Twitter account with around 800 developer followers (small, but engaged)
• About a year of hands-on experience using AI APIs for personal projects and client work
• A strong opinion about which platforms actually worked well in production What I didn't have: an audience that trusted me to recommend products. That's the gap. You can write all the affiliate content in the world, but if nobody's listening, you're shouting into the void. The fix was simple in theory, hard in practice. I needed to create content that solved real problems first, and let the affiliate recommendations follow naturally. More on that in a minute. # # Picking the Program (Here's the Commission Math) I applied to three affiliate programs during my first week. Two of them were one-and-done deals — you get paid once when someone signs up, then the relationship ends. No residual. No recurring. Just a flat fee and a wave goodbye. The third one was different. Global API runs on a recurring model:
• 15% commission on the first order
• 8% recurring commission every month after that
• 10% commission when someone upgrades to a premium tier
• And the platform itself aggregates 150+ AI models under one roof, which made the "what should I recommend?" question pretty straightforward Let me break this down for you in spreadsheet terms, because this is where it clicked for me. If someone signs up for a $20/month plan through my link:
• First month I earn: $3.00 (15% of $20)
• Every month after that, as long as they stay subscribed: $1.60 (8% of $20)
• If they upgrade to premium later: 10% on whatever that plan costs That's the difference between earning once and earning like a landlord. If a referral sticks around for 12 months, I'm earning roughly $22.20 total from that single signup, not just $3.00. And if they never cancel? That's $22+ every single year for doing nothing additional. The work is upfront. The income is passive-ish. That recurring structure is what sold me. I picked Global API as my primary recommendation and moved on. # # Month 1: The Slow Burn Week 1: I published my first comparison-style article — a piece based on actually using AI APIs in real projects, not just reading documentation. It was about 1,800 words, included working code samples, and recommended Global API naturally as my preferred platform for most developer use cases. I also cross-posted to Dev.to to expand my reach beyond my own blog. Week 2: Dev.to gave me my first visibility spike. The article picked up 340 views in its first week. My blog contributed another 120 views. So roughly 460 eyeballs. Not huge, but for a niche technical topic? I'll take it. But here's the part nobody talks about: I got exactly three clicks on my affiliate link that week. Three. And zero conversions. You know what three clicks and zero conversions means? It means I got paid exactly $0.00 for week one. Let me do the per-hour math on this article. I spent about six hours writing it (research, coding, editing, publishing, promoting). If I earned $0, my hourly rate was $0. Let that sink in. Week 3: Dev.to kept pushing the article, and views climbed to 520. Eight more people clicked my affiliate link. Still no paid conversions, but I got one signup. A signup is the bottom of the funnel. If they don't convert to paid, I earn nothing. But a signup means the content is at least doing its job — attracting interested people. Week 4: My first conversion happened on day 28. Someone signed up through my link and went straight to a Pro plan. My commission was $3.00. Month 1 totals from my Notion tracker:
• Articles published: 2
• Combined views: ~750
• Affiliate clicks: 14
• Signups: 2
• Paid conversions: 1
• Earnings: $3.00 first-order + $0.00 recurring
• Grand total: $3.00 Hourly rate: I logged about 14 hours of work in month 1 (writing, publishing, promoting, tweaking). That's roughly $0.21 per hour. Less than the cost of a gumball. But here's the thing I kept telling myself at my day job — every SaaS company loses money on month one. They make it back in month 12, month 24, month 36. My economics work the same way. Month 1 wasn't about the $3. It was about proving the funnel worked end to end. # # Month 2: Things Started Compounding Walking into month 2, I had one paying referral, two published articles, and 14 clicks of historical data showing people were at least interested enough to click. The question was whether the model could produce real income or if month 1 was a fluke. I set a target: $50 in total earnings by month-end. Ambitious? Probably. But I needed a number to chase. Week 5: I published article three — a case study about using AI APIs to build a client feature. This one performed differently. It got 280 views in the first week, but the click-through rate on the affiliate link was noticeably higher. Why? Because developers reading a case study are already in "I might build this" mode. They're not casually browsing. The intent is higher. Week 6: This is where compounding kicked in. My month 1 comparison article on Dev.to crossed 1,200 total views. Google started indexing it for some long-tail keywords. Clicks started coming in at 4-5 per day without me doing anything new. I got two more conversions to Pro plans this week, both from the original article. Week 7: I shipped my beginner's guide — a 2,200-word walkthrough for developers who'd never touched an AI API before. It took me nine hours to write, the most of any piece so far. But beginners convert at higher rates because they're actively looking for someone to tell them what to use. They don't have strong opinions yet. They trust recommendations more. Week 8: First recurring commission landed. $1.60 from my original month 1 referral, who renewed their subscription. I literally did nothing that week. I didn't write, didn't promote, didn't tweet. The money just appeared because the structure rewards retention. Month 2 totals:
• New articles: 3 (5 total)
• Combined views: ~2,100
• New affiliate clicks: 44
• New conversions: 3 (Pro plans)
• Recurring earnings: $1.60 (from month 1 referral)
• First-order earnings: ~$9.00 (3 × $3.00)
• Total: ~$10.60 added this month
• Cumulative earnings: $13.60 Hourly rate calculation: I spent about 22 hours on content and promotion in month 2. That's $0.48 per hour. Still brutal, but trending in the right direction. Up from $0.21. Progress. # # Month 3: The Tipping Point (Sort Of) By month 3, my portfolio was working for me. Five articles, indexed by Google, slowly accumulating views and backlinks. I was no longer relying on single posts to perform — the library effect was real. Old articles kept generating clicks while I wrote new ones. Week 9-10: My case study from month 2 hit a Dev.to trending page. Views exploded. I picked up 1,800 views in two weeks from that single piece. Affiliate clicks from this article alone: 31. One conversion. Week 11-12: I published two more articles — one focused on a specific AI API integration pattern, another on cost management for developers running AI in production. Both recommended Global API as my go-to for the 150+ models under one account (the "don't juggle five API keys" angle). By end of month 3:
• Articles published: 7 total
• Combined views across all content: ~6,400
• Total affiliate clicks: 187
• Total signups: 14
• Total paid conversions: 7
• Paying referrals still active: 6 (one cancelled) Month 3 earnings breakdown:
• First-order commissions: $18.00 (6 new conversions × $3.00 first-order)
• Recurring commissions: $8.00 (growing base of subscribers)
• Premium upgrade: $0 (no one upgraded yet, but that 10% rate is there for when they do)
• Month 3 total: $26.00 # # The Three-Month Spreadsheet View Let me put the full picture in one table because I know you want to see it. | Month | Hours Worked | Articles | Clicks | Conversions | Earnings | Hourly Rate | |-------|--------------|----------|--------|-------------|----------|-------------| | 1 | 14 | 2 | 14 | 1 | $3.00 | $0.21 | | 2 | 22 | 3 | 44 | 3 | $10.60 | $0.48 | | 3 | 19 | 2 | 129 | 3 | $26.00 | $1.37 | | Total | 55 | 7 | 187 | 7 | $39.60 | $0.72 | Yeah, $39.60 over three months. That's not a quitting-your-job number. I know that. But watch the trendline. Month 1 to month 3, my hourly rate went from $0.21 to $1.37. That's a 6.5x improvement in 90 days. And the recurring base kept growing. Each new conversion adds not just $3.00 to that month's total, but $1.60 every month going forward. My month 4 projection (just running the math): if I add 3-4 more conversions and keep my retention rate, I'll clear $40-50 in a single month. By month 6, assuming I keep publishing at this rate and retention holds, I'm looking at $70-90 monthly. That's not retirement money, but it's a meaningful side income from content I wrote once. # # What I'd Do Differently A few lessons from the trenches: Content over links. The articles where I led with solutions and let the affiliate link sit naturally performed 3-4x better than anything that felt like a thinly veiled promotion. Readers can smell an ad from a mile away. Dev.to is gold for technical content. I'm not exaggerating — Dev.to gave me more visibility than my own blog did in the first three months. Free distribution plus built-in audience. Use it. Track recurring like your life depends on it. The first-order commission is the carrot. The recurring commission is the actual business model. Every signup I lose costs me $1.60/month forever, not just $3.00 once. Don't quit your day job in month 2. I watched a few people in affiliate marketing forums flame out because they expected month-2 results. Build the base. Let it compound. The income is in months 6-18, not months 1-3. # # Should You Try This? If you're a developer with hands-on AI API experience and a small audience (or willingness to build one), the math genuinely works — but only if you treat it like a content business, not a get-rich-quick scheme. The creators I see succeeding in this space are the ones writing problem-solving tutorials, not SEO-spam comparison posts. For me, the deciding factor was the recurring