In 2019 I wrote an article on how companies should price their AI-enabled software. I focused on SaaS companies that were developing their own AI and highlighted pricing considerations as they worked to improve their models.
Since then, there has been a meteoric rise of third-party fundamental model providers like OpenAI, MosaicML and more. These “AI as a service” editors have enabled any SaaS player to integrate powerful AI into their application. This has created a mad rush to sprinkle AI pixie dust into the SaaS ecosystem. We’ve seen it among countless newly created startups and more established public companies.
The proliferation of this technology raises many questions, including how to deploy it safely, who will win (targeted startups or incumbents with existing distribution?), and more. One important area that hasn’t been discussed much yet: how it should be priced.
Below I outline a framework for how to think about AI pricing in your SaaS application. The space is changing rapidly, so I will update this thinking in future articles.
What differentiated value are your AI capabilities creating?
By definition, these fundamental models are available to all SaaS vendors, so how should you think about pricing what is, in effect, a product that you have integrated into your product? Start with the first principles: How much differentiated value created by this AI feature?
By integrating AI functionality into the flow of your larger platform, you save the user from leaving their flow to access the underlying model (ChatGPT, etc.). Keeping the user in context can be a powerful unlocker.
However, be honest with yourself about the value your AI is actually creating. Many AI features in SaaS today are getting a flood of initial kicks from curious users, but not seeing significant sustained adoption. Start by understanding retention and creating value.
SaaS companies should seek simplicity and adoption in the pricing of their AI features. It is a time of learning and iteration.
Next, ask yourself how differentiated your AI offerings are. If the majority of the value created by your AI functionality can be obtained by accessing ChatGPT directly, do not try to make a significant margin on this functionality. Reselling is not a sustainable value creation strategy (nor a differentiation strategy, although that is a topic for another article).
Even if you’re unable to charge much for your AI features today, they can create significant value by making your current product more valuable and possibly stickier. They can also be used to generate upsells to higher levels, which can lead to increased net dollar retention.
Over time, you can leverage initial features that today may be just a thin wrapper around a third-party model to create more differentiated value (see how below). When you get to this point, you can consider a higher value extractive pricing approach.
AI SaaS pricing is in its infancy