Pricing of AI products
Pricing any product is challenging, but given the stage of the curve we’re currently in, paywalling AI is nearly impossible for most…
Pricing any product is challenging, but given the stage of the curve we’re currently in, paywalling AI is nearly impossible for most companies.
As we consider Luzia’s future monetization strategy, I revisited my old 15.818 pricing notes to argue that paywalling AI-powered features is particularly complex and that finding a differentiation from non-AI features is key.
Two Basic Pricing Theories
Value-Based Pricing
For a product to be purchased, its economic value to consumers (EVC) must outweigh the value of the closest alternative and exceed its price. Mathematically, this can be expressed as:
To determine the appropriate price, the following must hold true:
Simply put, your price should be lower than the product differentiation you offer (what makes you unique), plus the cost of the comparable product. Let’s break this down with a couple of examples.
Example 1: Pricing CozyBear
Imagine you’ve created a new teddy bear, CozyBear, that has a built-in heating feature. Regular teddy bears cost $20. CozyBear keeps kids warm, saving $5 on extra blankets, and its lavender scent helps them sleep better, saving $4 on sleep aids.
Using the above equation, you could price CozyBear at a maximum of $29, which includes the value of the blanket (+$4), sleep aid (+$5), and the competitor’s price ($20).
Example 2: Pricing the iPhone
Let’s say the closest competitor to your beloved iPhone is a Samsung Galaxy. The Galaxy offers $600 worth of value through features like music, maps, and email. The iPhone adds $200 worth of value through a frictionless experience, integration with the Apple Ecosystem, and superior photography. However, it lacks the Galaxy’s waterproof certification, requiring a $50 case for summer photos.
In this scenario, a rational consumer would be willing to pay up to $750 for the iPhone.
Knowing the value of each variable in real life is nearly impossible, especially since the value is subjective and varies from person to person. However, at least in theory, EVC remains the gold standard for pricing.
A More Basic Pricing Model
When companies can’t accurately estimate value, they often resort to cost-based or competition-based pricing:
Cost-based pricing: Determine what it costs to serve the feature/product, add a margin, and that’s your price.
Competition-based pricing: Look at a competitor and price your product within the same range.
While intuitive, these approaches are risky and flawed for several reasons:
Understanding a product's fully loaded cost is challenging. Without accurate cost allocation, you might sell a product with a positive gross margin but still face financial trouble.
Just because you spend money producing something doesn’t mean someone is willing to pay. I could spend $200 importing ice cream from Antarctica in winter, but that doesn’t mean anyone will pay $201.
Pricing similarly to a competitor without clear differentiation can lead to a race to the bottom.
Pricing AI Products
Paywalling AI products is particularly challenging due to the uncertain value and rapid industry evolution. Let’s explore why.
You Should Not Price What You Can’t Measure
When Luzia went from 0 to over 1M users in less than 60 days, many competitors emerged with similar products. Some rushed to monetize by placing paywalls, offering limited free usage, or charging for “smarter” models. While this strategy initially gave profits to some of them, it was bound to fail for two reasons:
Unclear Value Proposition
Users were experimenting with Luzia-like assistants, so the value we were creating for them was unclear. The users and the companies were still determining what people would use these assistants for. We had thousands of people asking Luzia for jokes, poems, gardening advice, generating images — you name it. They came from Siri or Alexa, and they needed to completely change their mental model to what Luzia could make for them, and that took time.
It took us around six months and almost 20M users to start converging toward some clear use cases — education and companionship. It was then, and not before, that we began to improve and narrow the focus of our product to satisfy our audience and create defensibility. Over time, we became synonymous with AI for Gen Z in Latam and reached market penetration of over 50% in some countries — blowing up the big genAI names in volume (to this date, I believe we serve more queries in Latam than Bing).
If we had put a paywall in place too early, we would have alienated potential users before understanding where the real value was.
Going back to our equation, there were several problems. We (or the users, for that matter) did not know Value A, and our users were not clear who Product B was.
For those with short-term success, differential value (Value A — Value B) resulted from the hype created by a new eye-catching technology, and such a first-mover advantage gave you the ability to capture all the value. Their pricing equation looked something like this:
Over time, competitors and incumbents came up with similar solutions (featurization of products), and the lack of a clear use case reverted those gross revenues to 0. << Ey but those that got hundreds of thousands of revenue being solo companies, … well played >>
Rapid Industry Evolution
The AI industry is moving quickly. What’s valuable to early adopters today may be commoditized tomorrow. The fast pace and minimal differentiation across models diminish pricing power, leading to a race to the bottom.
Take the example of AI-powered transcription services. Initially, companies charged a premium for accuracy and speed. However, as more players entered the market with similar offerings, differentiation eroded, and pricing power diminished. Companies without solid branding or unique features struggled to maintain their position. Furthermore, competition among incumbents, with little differentiation, puts pressure on the margins. If Zoom, Meet, and Microsoft add these features, eventually, they won’t charge, and their margins will be impacted.
Some might argue that Luzia is in a similar spot, being “just” a wrapper over OpenAI. However, we anticipated the commoditization of the foundational layer and focused on brand and distribution, remaining technology-agnostic and betting on continuous improvement and cost reduction. The product will get better and cheaper, and everyone likes that. So far, so good.
The Role of Brand and Distribution in Maintaining Pricing Power
As AI models become commoditized, establishing a strong brand, effective distribution, and leveraging network effects are essential for maintaining pricing power. Similar to how Apple secures a premium through its brand strength, AI companies need to build a recognizable and trusted brand to differentiate themselves. This is akin to WhatsApp's dominance in messaging—a seemingly simple product, yet it holds a significant market position due to its brand and widespread use.
For Luzia, this means becoming synonymous with AI for Gen Z in Latin America — a position strengthened by our focus on education and companionship. Our users have naturally gravitated toward these two use cases, providing genuine value in their lives, the value that we now reinforce with our product and brand, creating a virtuous circle that has put Luzia among the top AI WW apps.
As AI products become more similar in functionality — because the underlying technology is commoditized — users’ emotional connection with a brand becomes a key driver of their willingness to pay.
Conclusion
Paywalling AI products today is highly challenging and demands a thorough understanding of the value being created — a task made difficult by the rapidly evolving technology. Premature monetization attempts can result in pricing mistakes, poor user retention, and, ultimately, failure. However, by effectively applying the Economic Value to Customer (EVC) framework and focusing on solid brand differentiation, AI companies can sidestep these common pitfalls, ensuring they maintain a competitive edge in an increasingly saturated market. Success hinges on patience, precision, and strategic branding.