Congratulations—your SaaS product is now powered by AI, and your users are seeing value. But beyond the excitement, your CFO has a new headache: every time a user uses an AI-powered feature, you get charged for the associated tokens.
AI introduces a variable cost to serve that scales with token consumption, not headcount. If you don’t adjust your pricing model, you’re subsidizing the costs of these tokens with your own gross margin. To navigate this, you must systematically re-evaluate your pricing and charging model to manage this new cost structure.
This article highlights heuristics to review your pricing model, price packaging and charging mechanisms in light of AI adoption. It assumes that you’re a profit-maximizing entity operating with a Good-Better-Best SaaS pricing architecture. Your circumstances are unique, and your strategic objectives or starting pricing architecture may be different.
1. Pricing Model: Price the Value of AI
The most advanced vendors are embedding AI directly into core workflows, as an intelligence layer. Is the value of your AI-powered features measurable? Is there an ROI for your customer?
When AI delivers a measurable and attributable result, anchor the price to the value delivered. Charge a value-based fee for every successfully resolved issue, like a resolved support ticket or a recovered fraudulent transaction. If the AI fails and the human in the loop resolves the issue, your customer doesn’t pay. This aligns incentives: the vendor is incentivized to make the AI more effective, and the customer pays a share of ROI. Outcome-based pricing is a powerful lever for monetization.
When AI does not deliver a measurable or attributable result, link the price to usage. AI becomes an add-on. Many leading B2B vendors charging on a per-seat basis have introduced AI usage thresholds. Once a customer hits a certain number of AI “credits” or “tasks,” they pay for expansion. This “pay-as-you-grow” approach offsets the vendor’s large language model token costs, while allowing customers to explore the new features at a low entry point.
2. Price Packaging: AI as an Upgrade Lever
In a traditional Good-Better-Best price packaging architecture, price-sensitive or entry-level users are offered a Good tier. Users requiring more advanced features are offered a Better or Best tier.
As a simple rule of thumb, do not offer AI in your Good tier. Reserve it for your Better or Best tiers. Putting AI in your Good tier is financially toxic. The incremental cost of tokens will drive your entry-level price point up, putting off price sensitive customers. Moreover, it will not be valued by users with basic needs. Reserving AI for higher tiers encourages upgrades and supports wallet share expansion.
3. Price Points: Offset the Token Bill
Your price points should increase to offset the new cost to serve from AI tokens, or your margin will suffer. You should monetize the value that AI creates for your customers using a value-based pricing lens, not just mark-up the tokens.
4. Discounting: Review Your Policy
By introducing AI, you’re adding a new variable cost – tokens – to your cost structure. Make sure that your discounting policy reflects this new reality. If you offer a blanket discount on an AI-enabled plan, you risk discounting below your own cost for tokens. You don’t want to lose money on every transaction!
5. Charging: Charge Tokens Upfront to Manage Cash Flow
To the extent that it is feasible, ask customers to buy AI credits upfront. By doing so, if you’re charged AI tokens in arrears by your LLM provider, you flip the cash flow in your favor. You also ensure that you aren’t footing the bill for a customer’s runaway consumption.
6. Trials: Limit Free Tokens to Manage Your Cost to Acquire Customers
Are you going to allow trial users to consume tokens? If you are, be deliberate about how many tokens are allowed in a trial period, to manage CAC. Trials should be credit-bound to limit your financial exposure.
Conclusion: Start Fixing Your Economics
AI didn’t add a feature—it rewrote your unit economics. If your pricing still treats AI as an expense to absorb rather than an asset to monetize, you’re already behind. The divide emerging in SaaS is between those who rebuild their pricing around value creation and those who quietly erode their margins with every token consumed. The market will reward those with the conviction to redefine what “value capture” means in the age of intelligent software.
Mores articles on Revenue
Receive exclusive Benchmarks and Research directly in your inbox Subscribe to our ASTERI+
Monthly Summary of ASTERI Insights and more Subscribe on Linkedin
