From Seats to Success: Building Flexible SaaS Pricing for AI Products (2025)

The software industry is experiencing a fundamental shift in how products are priced and sold. Traditional seat-based licensing models are giving way to usage and outcome-based strategies as AI transforms the value proposition: software’s worth now comes from insights and automation, rather than mere access to information. This evolution presents both opportunities and challenges for companies that seek to align their pricing with the actual value they deliver to customers.

AI is creating tremendous new business opportunities, but when I hear even Sam Altman asking his customers on social media what OpenAI should charge, I know we have a challenge to solve before AI can drive profitable businesses.

Seat-based pricing has dominated SaaS for years, offering simplicity and predictability. However, this model has critical limitations as AI transforms software capabilities:

With AI’s increased productivity, a few people can achieve what previously took dozens, which undermines the logic of charging by user count. More importantly, seat-based pricing fails to capture the differential value that users extract from your product. Does your 11th user derive the same value as your first or second user?

The Economic Value Curve: Why Usage-Based Models Win

The fundamental principle underlying this shift is alignment with the economic value curve. In usage-based models, pricing correlates directly with consumption patterns, creating a more accurate reflection of the value exchange between vendor and customer.

For companies embracing AI functionality, this alignment is essential. AI components introduce significant cost volatility — expenses can fluctuate dramatically based on usage patterns, model complexity and compute requirements. Creating pricing that accurately reflects these usage patterns requires considerable flexibility in both strategy and systems.

AI-Specific Pricing Challenges

AI introduces several distinct pricing complexities:

1. Model Cost Volatility

AI model costs can drop dramatically when new iterations are released, sometimes by orders of magnitude, or increase with more complex models designed for deeper reasoning.

2. Unpredictable Usage Patterns

Predicting customer consumption of AI features is challenging, especially since their users are still discovering how AI can add value to their workflows.

3. High and Variable COGS

AI-powered products often have substantial computing costs, and even though model prices may drop, consumption charges can escalate dramatically as usage increases.

4. Evolving ROI Perceptions

The market is still determining the true value of AI capabilities, so attempting to create a static pricing model contradicts how AI delivers value in practice.

As one customer explained to me during implementation, “In AI, model providers like OpenAI and Anthropic drop the price of a model 10X over nine months.” This volatility transforms pricing strategy from an annual consideration into an ongoing, iterative exercise.

Pricing in Flux: From Seats to Usage to Outcomes

The rapid changes in the AI market are pushing companies to rework how they price their products, but this evolution isn’t following a single path — it’s branching in multiple directions as organizations weigh seats, usage and outcomes to find the right balance between customer value and OpEx cost.

Many companies first transition to usage-based models, which measure and bill for the volume of activity rather than the number of users. For AI applications, this takes the form of:

1. Token-based pricing for generative AI applications.
2. Compute-time billing for inference engines.
3. API call charges based on resource consumption.

These resource-based pricing approaches introduce a new level of complexity into product pricing, requiring more sophisticated systems and strategies.

Usage-based approaches are increasingly being replaced by outcome-based pricing, which ties costs directly to measurable results. Consider customer experience platforms like Intercom and Zendesk, which can charge around 99 cents per successful ticket deflection. Each deflected ticket saves customers approximately $2.40 in support team costs, demonstrating clear ROI.

Given how the industry is very much in a transition phase, it’s not surprising to see all of these different approaches to figure out what’s considered valuable in an AI-world. But they all rely on having a technical infrastructure that can track, measure and bill for whatever unit of value makes the most sense for a company and its customers.

Navigating Operational Complexity

The shift toward usage and outcome-based pricing creates substantial challenges for pricing software products. Many companies find their pricing strategy constrained by existing tools rather than business objectives. I’ve witnessed companies make pricing decisions based on the limitations of internally developed tools, resulting in suboptimal outcomes for both the business and customers.

This misalignment typically manifests as:

1. Delayed product launches occur when billing becomes the development bottleneck.
2. Valuable engineering resources diverted to billing implementation.
3. Revenue leakage from inaccurate usage tracking.
4. Customer confusion due to lack of billing transparency.

For AI-powered products, these challenges are magnified by rapidly shifting costs. The ability to iterate rapidly on pricing becomes a critical competitive advantage.

Explaining Strategic Pricing Effectively

When the AI market changes almost daily, companies need to frequently evaluate their pricing so they can respond to changes in capabilities and costs. This is because more powerful models introduce new capabilities, potentially at a higher cost. Explaining these pricing changes to customers so they understand the additional value they’re receiving from a product is also important. When implementing price changes, provide advance notice and a clear rationale.

1. Balance Simplicity and Precision

Pricing should never be a spreadsheet exercise for customers. There needs to be a human who understands the pricing when they buy something and a human who understands the pricing when they sell it. Keep the pricing as simple as possible, and then simplify it even further.

A logarithmic discount curve might be mathematically optimal and “fair” to a company selling services, but that pricing strategy will fail if customers and sales teams don’t understand and can’t explain it. Field test pricing models with sales representatives and gather feedback from initial customer conversations before full deployment. Utilize tools that can model using actual historical usage data to determine how a price change will impact a particular customer, and can analyze different pricing models to best fit the customer’s usage patterns, and find the best fit between pricing and infrastructure costs.

2. Provide Options and Control

When transitioning existing customers, provide meaningful choices:

1. Option to lock in current pricing for a defined period.
2. Gradual migration paths to new models.
3. Clear documentation of how usage patterns translate to costs.

Giving customers a sense of control (and context) reduces resistance to change. One particularly effective approach is implementing “preview billing” — showing customers what costs would be under the new model while still charging them under the existing one.

3. Leverage Multiple Communication Channels

Communications about pricing changes should flow through multiple channels: company blogs, internal emails, product information materials, and social media (company handles and leadership). Ensure that customers understand the additional value provided with any pricing changes and can see how the investment will contribute to their business growth. Use these two-way channels to gauge the market response to pricing changes.

Connecting Communications and Technical Infrastructure

Effective pricing communication depends on supporting infrastructure that delivers accurate, timely information. The most transparent pricing communication fails if the underlying systems can’t reliably track, calculate, and report on usage patterns.

This connection between communication and technical capability is particularly crucial for AI-powered products, where usage patterns fluctuate dramatically and costs change rapidly. When a customer asks, “Why did my bill increase this month?” your team needs immediate access to granular usage data that clearly explains the change.

Customers will accept pricing changes when they understand them, but understanding requires visibility, and that depends entirely on the robustness of your technical foundations.

Technical Requirements for Modern Pricing Models

Building a powerful pricing model requires technical infrastructure that ensures profitable billing while maintaining the flexibility to adapt to market changes.

1. Data Capture: The Foundation

“You can’t bill for what you don’t track” is fundamental to modern pricing. Organizations should:

1. Capture more dimensions than current pricing models require.
2. Include relevant context metadata with each usage event.
3. Maintain data granularity without premature aggregation.

Effective usage tracking captures multiple dimensions of each interaction, such as:

1. Which endpoint was accessed?
2. What resources were consumed?
3. What business outcome resulted?

For AI-powered products, this dimensional approach is crucial. A company might start by charging for basic API calls but later shift to outcome-based billing. The right data infrastructure makes these pivots possible without significant reengineering efforts.

2. Systems Designed for Experimentation

Modern pricing infrastructure must support:

1. Simulation capabilities to test pricing against historical data.
2. Flexible mechanisms to implement new calculations quickly.
3. Operational resilience for retroactive adjustments.

Finance teams often need to backdate contracts by days or weeks. With continuous data streams, handling weeks of missed data becomes a complicated data engineering challenge that requires robust systems designed for such operations.

3. Iterative Implementation Approach

The most successful organizations approach pricing evolution as an ongoing journey:

1. Start with comprehensive usage measurement before changing models.
2. Implement preview billing to validate the impact before committing.
3. Pilot new approaches with new customers first.
4. Gather feedback continuously and adjust as needed.

Pricing as a Strategic Advantage

The pricing revolution in software isn’t merely about using different billing methods – I think it’s a core change to the value exchange between vendors and customers. Companies that succeed in this new paradigm will view pricing as an ongoing conversation rather than a static decision. They’ll deploy technical systems capable of adapting as rapidly as AI itself evolves, while maintaining communication channels that preserve trust through change.

The future of software monetization won’t be found in any single pricing model, but rather in the organizational capability to evolve pricing as continuously as products themselves — measuring, communicating, and adjusting in an endless cycle of refinement that keeps pace with the accelerating evolution of AI technology.