New trend in AI application pricing: Pay-per-use!

Many venture capitalists have discovered that AI startups are adopting a new business model - usage-based pricing, instead of the traditional user-based pricing model.

According to a summary by a US technology media outlet, Cresta, a generative AI startup, initially charged users based on a per-user basis but has now shifted to charging based on every conversation its AI tools assist contact center employees with.

Customer service company Intercom released an AI chatbot called Fin in March, pricing it at 99 cents per customer request it can handle, which is different from the company's core customer service product that charges on a per-user basis.

Hume AI, a research lab and AI startup focused on analyzing people's emotional changes using AI technology, tone, and facial expressions, has also started charging based on minutes, annotations, and word count.

Public information shows that usage-based pricing (UBP), also known as consumption-based pricing, allows customers to pay based on the actual usage of the product, with metrics measuring how customers derive value from the product.

Currently, UBP pricing is becoming increasingly popular in the Software-as-a-Service (SaaS) field, gradually replacing more traditional subscription and per-user pricing models. Due to its direct correlation between the price customers pay and the value they receive from the product, this pricing model is considered to be synonymous with value-based pricing.

Karthik Ramakrishnan, a partner at venture capital firm Institutional Venture Partners (IVP), believes that the usage-based pricing model can help AI startups align product pricing more closely with the actual value they provide, with the latter being measured by the time and workload they save for customers.

However, compared to the traditional per-user pricing model, pricing based on usage (also known as pay-per-use) may not be able to lock customers into packages that generate more predictable revenue streams. C3.ai, a publicly traded company focused on enterprise AI and AI application development, encountered challenges with fluctuating revenue and gross margins when transitioning to UBP pricing.

Currently, there are three main types of usage-based pricing models:

• Pay-as-you-go: Customers only pay for the actual usage or consumption, which is suitable for companies with fluctuating business needs.
• Per-unit pricing: Customers pay based on the usage of resources counted in units. This model is preferred by cloud providers that offer more granular services.
• Tiered pricing: Customers can choose the level that suits their needs, and if their usage exceeds the limit, they will be upgraded to the next tier with higher pricing. Usually, there is a free package available for initial use. Based on usage-based pricing, also known as "metered services," it is similar to the metered service model of purchasing electricity or water from utility companies in real life. This pricing model was initially favored by SaaS and Infrastructure as a Service (IaaS) cloud providers, allowing customers to explore how to use the service in a natural way without having to subscribe in advance, thus retaining customers.

The advantage of UBP pricing lies in the transparency of the pricing model, which makes it easier to directly link the customer's usage costs to the supplier's resource consumption. For users, they can start using the product at a relatively low cost, minimizing adoption resistance. For suppliers, allowing more users to access the product within the same account can generate more new use cases, and even encourage a group of users to share experiences with other potential users within the company or external organizations, thereby expanding the total addressable market (TAM).

On the downside, this pricing model relies on the ever-changing demands of customers, which may make it more difficult for suppliers to predict financial data and achieve sustainable recurring revenue, and may even harm long-term growth of the business. However, data shows that in the past five years, the adoption rate of UBP pricing in the B2B SaaS field has almost doubled, with three-fifths of companies using some form of UBP strategy.

Another venture capitalist partner at Menlo Ventures, Naomi Pilosof Ionita, pointed out that in addition to the need for faster strategies to prove value to potential customers due to the novelty of the product, AI startups may cause customers to eventually hire fewer employees if they improve employee efficiency. This means that the number of user seats that generate revenue for AI companies under the traditional subscription model will decrease.

All these reasons make AI startups more willing to try new pricing models. At the same time, in the current macroeconomic challenges, enterprise customers are increasingly laying off employees and cutting expenses, and it takes longer for them to make software purchasing decisions. Usage-based pricing may be more easily accepted by enterprises because it allows customers to flexibly adjust expenses over time.

There is also analysis pointing out that the rise and gradual popularization of UBP pricing are closely related to the characteristics of the technology itself:

• Automation: Software is increasingly automating manual processes. The more successful the product, the fewer user seats the customer needs, and pricing based on user seats cannot change with the value generated by automation.

• Artificial Intelligence: Artificial intelligence takes automation further, ultimately eliminating the need for an entire team to continuously perform tasks, which means monetization is no longer solely tied to human users of the product.

• API: For many rapidly growing software companies, the value lies in the API (the ability for software to directly communicate with other software) rather than the UI (user interface), and value can be seen without having users.