If you work in the SaaS industry, it is likely that you have participated in discussions at your company regarding how your customers can benefit from the increased value of your products infused with generative AI, large language models (LLMs), or custom AI/ML models. While strategizing your approach and creating the product roadmap, it is crucial to underline an important aspect, one that can be compared to the California Gold Rush. Just as you wouldn't show up to the gold rush without a shovel, it is important not to overlook the monetization aspect of your SaaS + AI. Incorporate it from the beginning and integrate the necessary measures right from the start, rather than considering it as an afterthought or a post-launch task. Two years ago, I discussed the inevitable shift towards metered pricing in the SaaS industry. At that time, the catalyst to propel this shift was unknown, but the fundamental thesis remained intact. It was unforeseen in 2021 that a particular form of AI would act as that catalyst. It is important to realize that what is required is not just a "pricing" change, but a change in the entire business model. Traditional SaaS pricing has been a relatively straightforward process with a simple per seat model and a price point set above the underlying costs to achieve the desired margins. A pricing change would involve modifying the charges, such as transitioning from $79 per user/month to $99 per user/month. On the other hand, a monetization model change represents a fundamental shift in how you charge.

With AI as a consumption vector, accurate metering and usage-based pricing models become inevitable. There are several excellent examples of companies already leveraging usage-based pricing to monetize AI, including OpenAI and other providers of foundational AI models and services. Additionally, companies like Twilio, Snap, Quizlet, Instacart, and Shopify integrate with these services to offer customer-facing tools. One of the challenges in monetizing generative AI is that the length of prompts and outputs varies, with larger prompts requiring more resources to process and vice versa. Moreover, the usage patterns of different customers can vary significantly, with some using the tool sparingly while others generate new text multiple times daily for weeks on end, resulting in a larger cost footprint. Any effective pricing model must take this variability into account and scale accordingly. Furthermore, services like ChatGPT itself operate on a usage-based model, meaning that any tools utilizing ChatGPT or similar models should also implement usage-based billing to accurately reflect the underlying costs of providing the service. To ensure fair and transparent pricing, as well as frictionless adoption and user growth, companies should consider implementing usage-based pricing. The elastic front-end usage and corresponding back-end costs make generative AI products ideal candidates for usage-based pricing. Here's how you can get started: companies can utilize prebuilt or trained models from various providers, and further train them with their custom datasets before incorporating them into their technology stack as features. To have complete visibility into usage costs and margins, each usage call made to the AI infrastructure (whether it's an API call or a direct one) should be metered to accurately understand the usage and underlying cost footprint.