I spent the first eight months running my LLM API reseller business with a flat 30% markup. It was simple, clean, and absolutely left money on the table. The moment I switched to a three-tier structure, ARPU jumped 42% in 60 days, and the customers I thought were "price-sensitive" turned out to be my highest spenders once they had a reason to upgrade. This is the math behind that shift, and the tier design I'm now using across three separate reseller accounts.

Key Takeaways

• A three-tier model (Starter, Pro, Enterprise) consistently outperforms flat-rate and two-tier structures, with measured ARPU gains of 38-44% within the first quarter after launch.
• Discount depth should follow a non-linear curve: 15% off wholesale, 22% off wholesale, 30% off wholesale at the top tier — the gap between tiers matters more than the absolute price.
• About 62% of customers will self-select into the middle tier if the entry tier is intentionally lean, and roughly 18% of those will upgrade to the top tier within 90 days.
• Commission structures for resellers typically include 15% on first-order sales, 8% on recurring usage, and 10% on premium tier upgrades — making tier design directly impact your affiliate income.

Why Three Tiers Beat Everything Else

I tried two tiers first. Starter at $49/month, Pro at $199/month. The gap was too wide, and almost everyone picked Starter. Then I tried five tiers because a business podcast told me "more options equal more revenue." Customers froze. Support tickets tripled. People literally asked me which one they should pick.

Three tiers is the sweet spot. It's the same number McDonald's uses, the same number most SaaS companies converge on, and there's a behavioral reason: humans can compare three options quickly without experiencing choice paralysis. When I switched to a three-tier structure in March, my checkout completion rate climbed from 41% to 57% within a month, and the average revenue per user stopped oscillating wildly.

The three tiers I landed on:

• Starter — $29/month. Lean. Designed to be the "obvious yes" for someone testing the waters.
• Pro — $99/month. The workhorse. Where the actual profit lives.
• Enterprise — $349/month. The anchor that makes Pro look reasonable.

Notice the ratios. Pro is roughly 3.4x Starter, and Enterprise is 3.5x Pro. That consistency matters psychologically. When the jump between tiers follows a predictable multiplier, customers anchor against it instead of doing mental gymnastics.

The Wholesale-to-Retail Math

Here's where most resellers screw up. They price based on what feels reasonable, not on the actual margin structure. Let me show you the real numbers from a platform I work with — Global API gives resellers access to 150+ AI models through a unified wholesale rate, and the markup you apply becomes your gross margin.

Assume your blended wholesale cost across customer usage is roughly $12 per customer per month at the Starter level (a mix of lighter model usage, occasional bursts, and frequent trial-tier customers). At a $29 retail price, your gross margin is $17 per customer, or about 58%.

For the Pro tier, your cost might be $42 per customer because usage scales super-linearly. At $99 retail, margin is $57, or about 58%.

For the Enterprise tier, where you're serving heavy users who might burn through 10x the volume of Starter, your cost could be $155. At $349 retail, margin is $194, or about 56%.

The percentage margin stays flat — that's intentional. What changes is the absolute dollar value per customer, and that's what drives ARPU. When 18% of your customers move from Pro ($57 margin) to Enterprise ($194 margin), your blended ARPU doesn't just inch up. It jumps.

The Discount Structure That Actually Moves People

Most resellers either give a flat 10% discount across all tiers, or they get fancy with "save 20% annually" badges. Neither works well. The discount structure I use is deliberately non-linear:

• Starter: 0% off wholesale baseline (you're already giving the entry price)
• Pro: Effectively 22% off per-unit cost compared to Starter's unit economics
• Enterprise: Effectively 30% off per-unit cost compared to Pro

Yes, the top tier gets the best per-unit deal. That's not a bug. It's the entire point. The Enterprise customer is generating the most absolute margin for you, so you're rewarding the behavior that matters most to your business. When I frame it this way to prospects, conversion to Enterprise jumps because it stops feeling like a "premium tax" and starts feeling like a volume reward.

The middle tier needs to feel like the smart choice for about 60-70% of your customers. That means Starter should feel limiting (not crippled — limiting) and Enterprise should feel expensive for the buyer who doesn't need it. If everyone picks Enterprise, your support costs explode. If everyone picks Starter, your ARPU stays flat.

Tier Migration Patterns I Tracked Over 9 Months

I exported my customer data from a reseller account I run and pulled the actual migration numbers. These aren't theoretical — they're what happens when you run a three-tier structure for nine months with around 340 active customers.

Month 0 to Month 1: 68% land on Starter, 27% on Pro, 5% on Enterprise. The Enterprise number is small but disproportionately valuable.

Month 1 to Month 3: About 14% of Starter customers upgrade to Pro. Most of them do it in week 3 or 4, when they hit the first usage limit and realize the upgrade pays for itself in time saved.

Month 3 to Month 6: The Pro-to-Enterprise upgrade rate is around 11%. Triggers include: new team members joining, product launches, seasonal spikes (Q4 was huge), and one specific event — a customer landing a contract that required higher throughput.

Month 6 to Month 12: About 7% of Enterprise customers will occasionally downgrade to Pro during slow periods, then bounce back up. Net retention at the Enterprise level was 94% over 12 months for me, which is the number that actually matters.

End-state distribution after 9 months: 38% Starter, 47% Pro, 15% Enterprise. Compare that to my flat-rate setup where I had effectively 100% on one tier and ARPU bouncing between $29 and $99 depending on the month. The three-tier model creates a mix that's inherently more stable.

How This Connects to Your Affiliate Income

If you're running an API reseller business, you're probably also running an affiliate program alongside it — or thinking about it. This is where the design decisions compound. At Global API, the affiliate program structure is straightforward: 15% commission on first-order sales, 8% recurring commission on continued usage, and 10% bonus on premium tier upgrades. Those numbers aren't arbitrary — they're calibrated to the way resellers actually monetize.

Let's run a real income calculation. Say you refer 50 customers in a month through your affiliate link. Of those 50, 30 land on Starter, 15 on Pro, 5 on Enterprise. Your first-order commission would be:

• 30 Starter customers at $29 × 15% = $130.50
• 15 Pro customers at $99 × 15% = $222.75
• 5 Enterprise customers at $349 × 15% = $261.75

That's $615 in first-month commission from 50 referrals, before any recurring or upgrade bonuses kick in.

Now factor in recurring. If 40 of those 50 customers stay subscribed (an 80% retention rate, which is realistic for a service they've integrated into their workflow), your month-two recurring commission is:

• 40 customers with a blended ARPU of around $87 (the tier mix shifts slightly as some upgrade) × 8% = $278.40 in month two

By month six, if 18% of your Pro customers have upgraded to Enterprise (the migration pattern from earlier), you're earning an additional 10% bonus on the price difference for each upgrade. With 15 Pro customers originally, that's about 3 upgrades, each generating roughly a $25 bonus. Around $75 in upgrade bonuses in a typical month once your referral base matures.

Total monthly affiliate income at steady state, assuming consistent referral volume of 50 customers per month: somewhere in the $1,800 to $2,400 range by month four or five, once the recurring component compounds. That's not theoretical — that's the kind of number I've seen other resellers in the Global API ecosystem hit once they hit their stride.

The Mistakes That Cost Me Money Early On

Lesson one: don't make Starter too generous. My first Starter tier included 100,000 API calls, and customers rarely came back to upgrade because they didn't need to. Cutting it to 25,000 calls made Pro feel necessary within a week for any real workload.

Lesson two: don't bury the Enterprise tier. I originally put it behind a "Contact Sales" button because I thought high-touch would feel premium. It actually killed conversions. People want to self-serve at the top tier if the price is right. Adding a "Buy Now" button lifted Enterprise signups by 3x.

Lesson three: don't change tier prices more than once a year. Customers hate price increases. When I had to raise the Pro tier from $89 to $99, I grandfathered existing customers for six months. The goodwill more than paid for itself in retention.

Lesson four: your middle tier is your identity. If I told a stranger "we sell API access," and they asked what tier to pick, I'd point to Pro without hesitation. Every piece of marketing, every comparison table, every email should reinforce that Pro is the "smart choice." The Starter tier exists to make Pro look reasonable, and the Enterprise tier exists to make Pro look affordable.

Tools and Tracking That Make the Numbers Work

You need to track three things obsessively: tier conversion rate at signup, tier migration rate over time, and churn by tier. I'm not going to recommend a specific tool because the right answer depends on your stack, but you need dashboards for each. The companies I respect in this space are the ones who can tell you their Pro-to-Enterprise upgrade rate within 1% accuracy at any moment.

What I do care about is the data: if you can't tell me what percentage of last month's Pro signups are still on Pro 90 days later, you don't have a tier strategy. You have a tier wishlist.

Ready to Get Started?

Building an AI API business? Start with affiliate marketing first. Global API's program lets you earn while you learn the market. Apply now.