AI products are not 85%-margin software
Classic SaaS carries near-zero marginal cost per user. An AI feature has real COGS on every action, which is why AI-native companies commonly report 50–70% gross margin instead of 80%+. Price and investor expectations both need to reflect that — a $9 unlimited plan on a $0.012-per-action feature is a subsidy, not a business model.
The heavy decile decides your pricing model
Usage is never normally distributed. The top 10% of users routinely drive 40–60% of total consumption. If your heavy cohort has negative margin, no amount of median-user profit fixes it at scale — those users are also the loudest advocates and the least likely to churn, so you compound the loss.
- • Model median and heavy users separately; never price off the average alone.
- • Cap unlimited plans with credits, fair-use limits, or a usage tier.
- • Route cheap steps to cheap models so heavy usage costs less per action.
- • Cache aggressively — heavy users often repeat similar requests.
Three pricing structures that survive AI COGS
Credit packs (usage visible, margin protected but friction added), hybrid seat + included credits with overage (the current market default), and outcome pricing where you charge per successful result. Pure unlimited-seat pricing only works when your cost per action is genuinely negligible relative to price.
Related guides
Long-form playbooks on the same topic, written by the RevenueLab editorial team.
LLM Token Costs in 2026: Pricing Every Model, Hidden Multipliers, and Margin Math
Input vs output token pricing across GPT, Claude, and Gemini, the context-window cost trap, how caching and batching cut bills 40–80%, and the real per-user margin most AI apps miss.
Read the guideSaaS Pricing Strategy: Per-Seat, Usage, Tiers, and the Hybrid Future
A framework for choosing a SaaS pricing model — when per-seat caps your growth, when usage-based makes revenue volatile, and how hybrid models stitch the two together.
Read the guideFAQ
What gross margin should an AI product target?
60–75% is the realistic band for an AI-native product today. Below 50% you'll struggle to fund sales and R&D; above 80% usually means the AI is a light garnish rather than the core value.
How do I stop heavy users from destroying margin?
Include a generous credit allowance in the seat price, then charge overage. This keeps the pricing page simple for the 90% who never hit the cap while making the heavy decile pay for what they consume.
Should I charge per seat or per usage?
Hybrid, in most cases: a seat price for predictable buyer budgeting plus included usage and metered overage. Pure usage pricing scares procurement; pure seat pricing exposes you to the heavy tail.
How do I estimate cost per action?
Take the tokens for a typical request — system prompt, retrieved context, user input, and output — and price them at your model's rates. Run it in the LLM inference or chatbot cost calculator, then add a 20–30% buffer for retries and failures.
Do I include support cost in COGS?
Yes. Support, hosting, and any human-in-the-loop review are all cost of revenue. Excluding them produces a flattering margin number that falls apart the moment you scale.
What if my margin is negative at launch?
That's survivable briefly if you have a clear path down the cost curve — model prices fall, caching improves, cheaper tiers arrive. It is not survivable as a strategy. Set a date by which margin must be positive and instrument usage now.
How this calculator is built
Independently maintained
Written by Sam Doshi and the RevenueLab editorial team. We don't sell the data feeds this tool is built on.
Sourced from primary data
Benchmarks come from public AdSense / Stripe / IRS disclosures and reader-submitted data — never third-party "$X per view" claims. Full methodology.
Last editorial review
Reviewed on a rolling quarterly cycle. Dated reviews are published on the methodology record for each calculator.
Editorial standards
See our editorial policy and disclaimer. Results are estimates, not advice.