Pricing An AI Product | SBI Growth

What about using a secondary value metric?

To address the limitations of add-on pricing, companies can adopt a secondary value metric that captures the additional value provided by the AI features. This approach involves identifying specific metrics that reflect the usage and value derived from the AI functionalities. By tying the pricing to these metrics, companies can better align their revenue with the value experienced by users.

Grammarly provides an excellent example of this approach. While its core product is priced per user, the company differentiates pricing based on the volume of AI usage. Grammarly's plans offer different levels of access to AI functionalities, such as advanced writing suggestions and plagiarism checks. 

By doing so, Grammarly ensures that users who heavily rely on these AI features pay more, thus aligning the pricing with the value derived from the product.

AI as the core product

For companies where AI is the product itself, creating a new customer journey, nailing the value metric for pricing is critical. An effective value metric should be:

1. Value-Aligned: Pricing should reflect the value experienced by the customer.
2. Growth-Aligned: As customers use the product more, the metric should scale.
3. Measurable and Implementable: The metric must be objective and understandable to both the provider and consumer.

Finding the right value metric for complex AI products is challenging but essential. 

ChatGPT, currently priced per user, faces issues similar to Notion AI—treating all users equally, regardless of usage. Exploring alternatives like pricing per question or word count has limitations. Instead, a more nuanced approach might be needed.

MidJourney's innovative pricing

MidJourney, a natural language image generator, exemplifies innovative AI pricing. By deeply analyzing user behavior, MidJourney identified fast GPU time and maximum concurrent jobs as key value metrics. 

These metrics differentiate high-value users from casual ones, allowing for effective monetization. This approach highlights the importance of understanding what drives value for different user segments and developing pricing strategies accordingly.

Common questions about pricing for AI

Q: How would you recommend assessing willingness to pay? Considering tools, data, etc.?

A: It depends on where you are in the maturity of your product and your business. A good starting point, if you have existing customers, is to start talking to those customers, asking questions that are a bit of a workaround to understand willingness to pay. We often use a specific methodology where we're never asking someone directly how much they would pay for the product, but we're trying to understand the range of their willingness to pay so we can start to formulate a strategy. If we want to monetize and prioritize revenue, what price point can we consider? On the flip side, if we're going for acquisition, what's on the low end? Talking to customers is the foundation of any good pricing strategy. If you pair that with product usage data to understand value drivers, you can identify what to lean into for upgrading and cross-selling customers. If you don't have existing customers, for a new product, there’s desktop research and competitive analysis you can do, but you want to limit the barrier to adoption to gather that data, which will be really useful.

Q: How will the higher operating costs of generative AI systems impact pricing strategy?

A: This is where we get into the discussion about pricing being a mix of art and science—the science being looking at data and running a solvent business, and the art being mapping your product to a description of value and willingness to pay. Both have to work in tandem.

On one hand, it involves going after lookalike buyers and testing willingness to pay. The other is defining the outcomes you're aiming for and reverse engineering your strategy. Some products should be loss leaders, optimizing for volume play and getting customers in the door. Others, especially if they add significant value to your customer base, can be sold as an add-on, boosting metrics like NDR, ARPU, and ACV, which is healthy for margins. There are many use cases for these new AI systems, and it largely comes down to defining the outcomes you want.

Q: Do you think use case-based pricing or verticalized pricing will become more prevalent in AI?

A: It could become more prevalent. The challenge is supporting many different SKUs and implementing a complex pricing structure while keeping track of all that. Also, having clear use cases for your AI—being able to identify what those six or seven use cases are—is essential. Often, when exploring use case pricing, it gets blurry because the use cases overlap quite a bit, making it hard to articulate distinct value and ascribe a price point to it. For products selling primarily to SMB or mid-market companies, use case-based pricing might not make sense due to the volume play and overlapping use cases. For enterprise solutions, it might be worth investing in specific use cases due to the larger contracts and customization involved.

Q: How should AI-first companies price their products when going up against an incumbent?

A: Pricing can be a huge differentiator in this space. Staying value-aligned and usage-oriented can help separate you from an incumbent. However, it’s crucial to ensure that your pricing is comparable because customers will care about whether you're cheaper or more expensive. Having a unique pricing model is beneficial, but it’s also important that you can explain in sales conversations how your pricing compares to the incumbent. For disruptors going up against large companies giving away AI functionality, it’s crucial to have a clear value proposition and a pricing model that's simple, value-aligned, and measurable.

Q: How can AI companies improve customer understanding of the value they get from an AI product?

A: Talking to customers is a fundamental step. Using product analytics tools like Hotjar can help evaluate how people are using the product and identify differences in behavior and value experience. It's about reverse engineering the best proxies for the ultimate outcomes customers seek from your product. Additionally, using your own product to understand the customer journey and where value points are hit is a great way to improve understanding. Ensuring that customers are trained to understand key value metrics, through in-product notifications for example, can also help.