Pricing and Margins

The "$10/Month" Problem

Walk into any cohort of first-time SaaS builders and ask how they set their price. Most will say some version of: "I looked at what competitors charge and went slightly lower" or "I asked friends what they'd pay" or "It felt right."

These are not pricing strategies. They're guesses with a dollar sign attached.

The problem isn't that these founders are being careless — it's that they're optimizing for the wrong thing. They're trying to find a price customers will accept, instead of finding a price that makes the business viable.

Those are different problems, and they require different math.

Working Backwards From Cost

The only sustainable pricing approach starts with your cost structure and works backwards to a price. Not forwards from what competitors charge.

The formula is simple:

Price = Cost per user ÷ (1 − Target gross margin)

If your cost per user is $7.50/month (API + infrastructure + auth + support) and you want 70% gross margin:

$7.50 ÷ (1 − 0.70) = $7.50 ÷ 0.30 = $25/month minimum

That's your floor — not your target, your floor. The lowest price at which the business is viable at your target margin. You can price higher. You cannot sustainably price lower.

What Gross Margin Actually Means for AI SaaS

Gross margin is the percentage of revenue left after you pay the direct costs of delivering the product. It's what funds everything else: sales, marketing, product development, your salary, your runway.

Traditional SaaS typically achieves 80–85% gross margins because infrastructure costs are mostly fixed and scale slowly. A 1,000-user SaaS on Vercel pays roughly the same server bill as a 100-user SaaS.

AI SaaS is structurally different. LLM inference is a direct variable cost — every user request costs money. This creates a floor on gross margins that doesn't exist in traditional software.

In practice: well-run AI SaaS products achieve 60–70% gross margins. Below 50% is the danger zone. Below 30%, growth accelerates your losses.

The Four Pricing Mistakes

Mistake 1: Underpricing to acquire users.

"I'll start low and raise prices once I have traction." This creates a customer base of price-sensitive users who signed up because you were cheap. When you raise prices to a sustainable level, they churn. You've spent months acquiring users who valued the discount, not the product.

Price at a viable level from day one. Acquire users who value the outcome.

Mistake 2: Copying competitors without knowing their unit economics.

If a competitor charges $15/month, you don't know what their costs are. Maybe they built on a cheaper model. Maybe they serve a different usage pattern. Maybe they're losing money and don't know it yet.

Pricing decisions must come from your cost structure, not from a competitor's landing page.

Mistake 3: Ignoring power users.

Usage-based costs don't distribute evenly. The top 10% of your users often generate 40–50% of your API costs — at the same price as the median user. If your pricing is flat, power users are subsidized by light users.

Build usage tiers that reflect actual cost tiers. Cap usage per plan or charge overages. "Unlimited" is only sustainable if your margins at median usage fund the excess cost of heavy users.

Mistake 4: Treating the free tier as a marketing cost rather than a product cost.

In traditional SaaS, free users cost almost nothing to serve — marginal infrastructure is near zero. In AI SaaS, every free user who makes AI requests generates API costs.

Know your cost per free user per month. Know your free-to-paid conversion rate. Calculate whether the free tier is an efficient acquisition channel or a subsidy program. If you don't have those numbers, don't launch a free tier.

What a Healthy Pricing Structure Looks Like

Start with three tiers based on actual usage bands:

Starter: Covers light usage, achieves 50–60% gross margin. Designed for customers who need core value without high-frequency use.

Pro: Covers medium usage, achieves 65–75% gross margin. This is where most customers should land. The pricing here funds your growth.

Power: Covers heavy usage, includes explicit usage caps or overages. Achieves 60–70% gross margin even at peak usage. Does not subsidize the 5% of users who generate 30% of your API costs.

Each tier's price is set by the cost formula, not by what sounds reasonable.

The Conversation You Need to Have With Yourself

Before you publish a price:

1. What is my cost per user per month at each usage tier?
2. What gross margin do I need to run a healthy business?
3. What does that formula give me as a minimum price?
4. Is that price one my target customers will pay?

If the answer to question 4 is no, you have a positioning problem or a cost problem — not a pricing problem. You need to either justify the price with stronger outcome messaging (the “Selling Outcomes, Not Features” lesson), reduce your costs by optimizing prompts or switching models, or reconsider whether you're solving a problem people will actually pay for.

Pricing that works starts with honest math. Everything else is creative presentation on top of that foundation.

For more on the business fundamentals of AI products, jeremyknox.ai has ongoing coverage of the builder ecosystem.