The Future of AI Pricing: Adapting to Value, Usage, and Market Needs

The Future of AI Pricing: Adapting to Value, Usage, and Market Needs

AI-native businesses are rapidly transforming industries, but pricing strategies are not developing as quickly. The old SaaS playbook which relied on per-seated or tiered subscriptions does not make sense when AI is replacing human work. As we all try to keep up with the pace of AI innovation, it is worth revisiting pricing strategies and considering how they should be updated for AI-native businesses. I believe pricing will no longer be a one-size-fits-all approach but rather a blend of models which takes into account the end market, ROI alignment, and trust with the end customer.

This post will explore emerging AI pricing models, break down their pros and cons, and share insights on which strategies are gaining traction in today's market.

Where Outcome-Based Pricing Wins and Where It Falls Short

Outcome-based pricing is aligning a company’s revenue with the business impact on its customer. The best example of this is in customer support where AI-native companies generate revenue per ticket resolution. Major players in the customer support space – such as Zendesk, Intercom, and Front – have introduced hybrid pricing structures that combine per-seat with a charge per ticket resolution. This model works well when it’s easy to compare AI performance to human labor and there are clear ROI metrics.

Another area where outcome-based pricing is gaining traction is AI-native vertical applications. For example, Thoughtful AI is automating aspects of the revenue cycle management process and charging based on measurable outcomes which they outline with their customers.

However, there are some drawbacks to this model that must be considered:‍

1. ‍Misaligned Incentives: An AI agent could close tickets quickly but leave customers unhappy, leading to repeat issues, or it could optimize for fast resolutions rather than solving problems at the root. This enhances the need for companies to partner with a reliable software provider that they trust and have tested.
   ‍
2. ‍Budget Unpredictability: If an AI tool resolves 10times more tickets than expected, it is great for efficiency, but the cost could go much above a company’s budget. The trade-off between cost and efficiency can be worth it if the business sees real returns, but it’s still a risk. Also, not every industry can afford outcome-based pricing, even if it delivers better results. Sectors like education, public safety, and government often have fixed annual budgets. Even if an AI tool delivers massive efficiency gains, a public school district or a local government agency may not have the flexibility to pay more based on outcomes. In these cases, predictable pricing models often win out, even if they are less aligned with value.
   ‍
3. Falling Costs: Right now, some outcome-based pricing models anchor their rates to human labor costs, but this approach won’t hold up long-term. As AI models become more efficient and compute costs drop, the cost to support an AI agent is going to decline significantly. Combined with the fact that more AI agents will enter the market, competition will force prices down even further. This means that the rates for outcome-based pricing will quickly get out of sync with the actual cost of delivering outcomes.

The Role of Usage-Based Pricing and Credit Models

A middle-ground approach between outcome-based and traditional subscription pricing is usage-based pricing, which is often structured as pre-purchased credits. This model can provide budget predictability for customers while still aligning with usage.

A great example of this is Clay, a sales intelligence software recently valued at $1.25B. Clay has implemented a credit-based pricing model where customers pre-pay for credits that can be spent on AI-powered workflows. This method helps prevent unexpected cost surges while ensuring vendor revenue remains aligned with customer usage in a sustainable manner.

Is Per-Seat Pricing Dead? Not Necessarily

While some in the industry are quick to declare per-seat pricing as outdated, it isn’t going away entirely. Its effectiveness depends on whether the AI is replacing human labor or assisting it.

• If AI can reduce a company’s headcount, charging per seat will cannibalize an AI-native company’s revenue. Per the examples above, an AI-powered customer service tool that resolves inquiries without human agents will drive down the number of human agents needed and, with them, the vendor’s revenue. Charging per seat in that scenario becomes self-defeating.
  ‍
• If AI enhances human performance rather than replacing it, per-seat pricing can still make sense. For example, Eleos Health reduces the time therapists spend on documentation by more than 70%. In this case, charging per therapist makes sense as the AI acts like a power tool for the therapist, not a replacement.

Per-seat pricing isn't dead, but rather it needs to fit the archetype of the role it plays.

Implementing the Right Pricing Strategy

If a company decides against per-seat pricing, the next challenge is how to implement a more flexible model. There are two main options: buy an existing pricing product or build a custom one.

‍

• ‍Buying a pricing tool: Companies such as Orb, Metronome, and m3ter provide purpose-built pricing infrastructure solutions designed for SaaS and AI-native businesses. These platforms automate real-time usage tracking, metering, and billing, reducing the need for companies to build these capabilities in-house. By leveraging these tools, businesses can quickly launch flexible pricing models, ensuring accurate billing while minimizing engineering overhead. This accelerates go-to-market strategies and allows teams to focus on their core product rather than complex pricing mechanics.
  ‍
• Building in-house: Developing a custom pricing system requires engineering resources and product integration. AI-native companies must create their own billing logic, implement real-time usage tracking, and manage the complexity of hybrid pricing structures. While building in-house offers maximum flexibility and customization, it is costly and time-consuming. Companies that choose this route must dedicate engineers to not only develop the system but to also maintain it.

Choosing between buy vs. build ultimately depends on the company's technical resources and flexibility needs.

Pricing Should Be a Conversation, Not a Contract

Ultimately, there is no single pricing model that works for everyone. It really depends on what the customer is trying to solve for, whether that is replacing humans or assisting them, how clearly ROI can be measured, and what their budget allows. A lot of companies are testing hybrid models to address these challenges, but it is still unclear what will win.

What I do believe is that testing to understand value for each customer, such as how Help Scout uses a three-month trailing average, and building pricing around that makes the most sense. Offering free trials or pilot programs during the sales process is one of the best ways to test value alignment. This approach lets customers see how much more efficient they become with AI while helping vendors collect real usage data to design pricing models that match business outcomes.

At the end of the day, customized pricing that aligns with value delivered is what will drive success. The companies that win will be the ones that make pricing a conversation – rather than a contract – built on customer data and real value delivered.