Generative pricing for AI is a blend of dynamic pricing and value based pricing
Steven Forth is a Managing Partner at Ibbaka. Connect on LinkedIn
Generative pricing is an emerging pricing methodology designed to meet the needs of generative AI companies, especially companies in what Bessemer Capital calls vertical AI. It has been designed to address key characteristics of generative AI applications. Second-generation generative AI applications have a number of characteristics that need to be factored into pricing and packaging design. Ask:
“Which of these are relevant to the application I am taking to market?”
Characteristics of generative AI applications impacting pricing
Dynamic resource requirements
These applications often have variable and unpredictable resource needs, especially in terms of compute power and GPU usage.
Pricing models need to account for fluctuating costs associated with running large language models and other AI systems.
Automation capabilities
Generative AI can potentially automate entire functions or jobs, making traditional per-seat pricing models less relevant.
Pricing needs to capture the value of automation and efficiency gains.
Exponential value creation
The value provided can grow exponentially as these systems take on more complex tasks.
Outcome-based or value-based pricing becomes more important to align costs with benefits.
Vertical specialization
Many generative AI applications are tailored for specific industries or use cases.
Pricing models may need to be customized for different verticals to reflect specialized value.
API-driven usage
Many generative AI capabilities are accessed via APIs, enabling more granular usage-based pricing.
Token-based or call-based pricing is common for API access.
Performance and quality considerations
Pricing may need to account for differences in output quality or model performance.
Premium pricing for higher quality or more capable models is common.
Integration and infrastructure costs
Pricing often needs to factor in costs beyond just the AI, including integration, data transfer, storage, etc.
Scalability requirements
Pricing models need to accommodate rapid scaling of usage as adoption grows.
Support and ongoing improvement
Costs for technical support, model updates, and continued development may be factored into pricing.
That is quite a list. Not everything on this list is relevant to every application, so focus on the 3-5 that will have the most impact on your packaging and pricing.
Limitations to current approaches to pricing for generative AI apps
Current approaches to pricing are not up to snuff for pricing this new class of applications.
Cost-Plus Pricing
Cost-plus pricing is particularly challenging for generative AI applications for several reasons:
The cost structure of AI applications is complex and dynamic, with costs varying based on usage patterns, model complexity, and computing resources required.
Fixed costs like model development are high, while marginal costs can be low, making it difficult to determine an appropriate markup.
Cost-plus fails to capture the value created for customers, which can far exceed the costs of providing the service.
Willingness to Pay and Dynamic Pricing
While willingness-to-pay and dynamic pricing approaches can be more responsive than static models, they have limitations:
They may not fully account for the rapidly changing configurations and capabilities of generative AI applications.
These models often lack transparency, which can be problematic for customer acceptance of AI-driven pricing.
They focus on extracting maximum revenue rather than aligning price with value delivered.
Value-Based Pricing
Traditional value-based pricing comes closer to addressing the needs of generative AI applications, but still has shortcomings:
Static value models may not keep pace with the rapidly evolving capabilities of AI applications.
They may not account for the composable nature of modern AI applications, where configurations change dynamically.
Traditional approaches to quantifying and communicating value may be too slow for real-time AI interactions.
Market Pricing
Market pricing faces challenges with generative AI due to:
The rapidly evolving nature of the technology makes market comparisons difficult.
The highly differentiated and customizable nature of many AI applications.
The potential for AI to create entirely new categories of products and services without clear market analogs.
The need for generative pricing
Given these shortcomings how do we move forward? This is a design problem. How do we design a pricing process that is fit for the new potentials opened by generative AI applications?
The design brief for this new approach to pricing derives from the unique characteristics of B2B generative AI applications and the shortcomings of existing pricing methodologies. Key considerations are …
Dynamic configuration
Conversational nature
Content rich nature of these applications
Data dependencies and use of synthetic data
Higher operational costs … lower development and support costs
After exploring many different approaches, and observing what is happening in the market (Michael Mansard at Zuora has been doing some compelling research here or read a summary from Perplexity) we decided that rather than come up with a completely new approach we could combine existing approaches is a concept blend or mash up.
We leaned on dynamic pricing and value-based pricing to begin developing generative pricing.
What is generative pricing? A concept blend of dynamic and value-based pricing
A concept blend of dynamic and value-based pricing to create a new approach to pricing - generative pricing
Concept blending is a compelling way to drive innovation. See Some innovation patterns from concept blending. Basically, within the context of a domain, one takes two different approaches or sets of concepts and to create something new.
For those who want to go deep into this see Turner and Fauconnier Blending as a central process of grammar. Generative AI will be a powerful way to generate, explore, and mange concept blends. The above concept blend was developed with extensive use of generative AI.
Generative AI itself is a powerful way to work with concept blending. Here is how we used Perplexity to help create the blend that is generative pricing. A careful reader will notice that we did not use the generative AI output ‘as is’ but used it to help spur and organize our own thinking, we used it as a thinking partner, not as a substitute for our own thinking.
Key aspects of generative pricing (an evolving perspective)
Customer-centric flexibility
Pricing for generative AI applications needs to respond to the customer’s needs, these include predictability, risk reduction, pricing transparency, and value to customer (V2C).
Model-driven value optimization
Generative pricing will require models, it inherits this from value-based pricing that relies on formal value models and from dynamic pricing, that is fed by data.
Personalized value stories
Pricing, and even value, can be dry topics that lack broad appeal. The solution is value stories. Generative AI applications need to be able to explain the value they provide and customize these stories for each customer.
Real-time adjustments
Generative AI applications will be able to reconfigure based on the customer’s current needs. This will put extreme pressure on pricing and packaging models which must be able to adapt in real time.
Multifactor differentiation
Economic value has two components, commodity value (the value that can be provided by many different alternatives) and differentiation value (the that is unique to the solution). Generative AI applications will need to provide different forms of value for each configuration, the configurations being determined by the customer’s needs.
More on Generative Pricing
Book a demo today and discover how our unique B2B SaaS solution can increase sales, improve retention, and drive growth.
