Steps on the path to generative pricing

Steven Forth is a Managing Partner at Ibbaka. See his Skill Profile on Ibbaka Talio.

At Ibbaka, we have been doing a lot of research into how the next generation of generative AI applications will be priced. We concluded that a new approach to pricing is needed, what we are calling “generative pricing.”

Here are links to our first three posts on generative pricing.

Generative pricing includes an approach that is more broadly relevant to many pricing scenarios. It supports a pricing process that helps pricing adapt as solutions mature and products evolve. One that makes it easy to align pricing to different packages of features. This is important …

  • During value based negotiation where functionality is added or subtracted in response to the buyer’s willingness to pay

  • When exploring what functionality to include in each package to set a price for the package

  • As part of product development, to understand what functionality will deliver the most value

The 5-Step Generative Pricing Process

It is still very early days for generative pricing, as it is for generative AI generally. That said, we know enough about second-generation generative AI applications to get started. Based on work we have done with a number of B2B SaaS companies working on new applications we have found that the following works.

  1. Build a value model

  2. Manage functionality using feature flags

  3. Map features to value

  4. Design pricing that lets you change pricing as you switch features on and off

  5. Connect pricing to configuration management

Build a Value Model

As we showed in The uses of value models, the value model sits at the center of the customer experience, connecting all the different pieces together. The value model, by which we mean an EVE or Economic Value Estimation style model as introduced by Tom Nagle and his collaborators in The Strategy and Tactics of Pricing, is the foundation of value based pricing. If you are not using a formal value model you are not doing value based pricing. Willingness to Pay (WTP) is not a substitute.

To build a value model one has to be able to answer some key questions:

  • Who gets value?

  • How do they get value?

  • How else could they get that value?

  • How much value do they get?

The value model answers these questions.

In generative pricing, one must go one step further and be able to connect value to functionality and features.

Manage Functionality Using Feature Flags

Software applications are composed of many discrete features. Some of these must be used together to complete a value path, others can be used independently. In many cases, there is more than one implementation of each feature (alternatives exist).

One way features are managed inside applications is with feature flags. A feature flag, also known as a feature toggle or feature switch, is a technique in software development that enables or disables a feature during runtime using a condition within the code. As the feature can be turned on or off during runtime it makes for dynamically configured software.

From Perplexity

Key Aspects of Software Features

  • Functionality: Features represent specific functions that a software application can perform. For example, in a word processing application, the ability to generate and modify documents is a feature.

  • User Value: Features are designed to meet the needs or solve the problems of users, thereby enhancing the user experience and satisfaction.

  • Differentiation: Features help differentiate a software product from its competitors by offering unique capabilities or enhancements that appeal to the target audience.

  • Configuration and Design: In feature-oriented software development (FOSD), features are used to describe the variability of a software product line and are implemented to satisfy specific requirements.

In generative pricing, we are concerned with the features that provide direct value to users, with a focus on the differentiated features (those that other software does not provide).

Map Features to Value

Feature to Value Mapping is the key to generative pricing. It is where the software’s functionality and the value it delivers get connected. There are several ways to do this, but the simplest is to generate a table with configurable features on one axis and value drivers on the other axis, as is shown above. For more complex software where there are a lot of dependencies between features one can construct a DSM (Design Structure Matrix) and put the value drivers in the cells.

Pricing Model

You can’t execute on generative pricing without a formal pricing model. Price books, tear sheets, pricing buried in other code … none of these will cut it for generative pricing, where you need to have a model that can be connected to and be driven by value-based configuration.

What makes something a pricing model?

A pricing model is a set of algorithms that sets a price for a configuration. The price can vary based on usage and more importantly the value of the usage. The pricing model includes:

  • A packaging model (how different sets of features, functions, and value paths can be combined to create value for buyers and users)

  • Pricing metrics (a pricing metric is the unit of consumption for which a buyer pays)

  • Pricing equations that integrate variables to specify a value for a pricing metric

  • Scaling algorithms, that adjust the price based on scale

Optionally, pricing models can also include

  • Discounting rules, that guide sales and deal desks on how large a discount can be provided and how that discount is earned or changes over time

  • Price adjustors, that automatically increase prices over time or when certain conditions occur

Configuration Management

Generative AI is making it much easier to configure applications. The old approach to configuration management was to use constraint based reasoning to limit the possible configurations to those that could actually be delivered. Generative AI flips this. Configuration becomes generative and the goal of configuration management changes.

Instead of constraining sales on what it can sell, generative AI configuration management guides the buyer on how to get the most value.

By connecting value to features and leveraging this mapping in configuration management the basic rules of the game are changed.

How the parts fit together

Generative pricing depends on the different parts of the system talking to each other. A systems approach is needed. The graph for generative pricing is cyclic, with feedback loops built in.

Value Model → Pricing Model

The value model determines the pricing model. Inputs to the value model are reused in the pricing model to construct pricing metrics. This is the key to value based pricing.

Value Model → Feature Value Mapping

The value model provides the value drivers that are one dimension of feature-value mapping.

Feature Flag → Feature Value Mapping

Defining features using feature flags is another key to generative pricing. These features are the other dimensions used in feature-value mapping.

Feature Value Mapping → Configuration Management

The feature value mapping is the key input into configuration management. It provides the structure and data needed for value based configuration.

Feature Value Mapping → Value Model

Understanding the value of a feature feeds back into the value model. The mapping process can lead to insights into existing and possibly new value drivers.

Configuration Management → Pricing Model

Configuration determines pricing. In a very real way, the configuration model configures pricing as well as the application.

Configuration Management → Feature Flags
The configuration management is able to switch feature flags on and off, determining the configuration.

Conclusion

The generative pricing approach can be applied today and is not limited to second-generation generative AI applications.

The benefits of this approach include

  • Explicit connections between configuration, value, and price

  • Adaptive pricing that responds to the value provided by the configuration sold and delivered

  • A systems approach with clear feedback loops between the different parts of the configuration - value - pricing system

Together these support a new approach to pricing optimization, one that cannot be delivered though dynamic pricing optimization.

Read other posts on generative pricing

Previous
Previous

Comparing the Pricing Pages of AiSDR, Reply.io, and Smartlead.ai

Next
Next

Understanding how generative AI SDRs are revolutionizing sales development