Estd 4.22.2023
AI in SaaS
How AI Is Forcing ARR to Play Atlas—and Why It Can’t Bear the Weight Forever
Read More
.png)
Monetizing GenAI
While some GenAI applications have indeed changed the game, there are far more AI products that still don’t add any real value and at the same time have highly variable cost structures, which creates a boom-or-bust effect in such products.
Unlike traditional software, which is WYSIWYG (what you see is what you get) where value is predictable and deterministic, AI models deliver probabilistic outputs. This makes demonstrating AI value a challenge, as the results are not guaranteed to follow a consistent pattern.
This is explained in Fig 1, with traditional SaaS software, the value is often simple—the output is deterministic. But with GenAI, the situation is different. The output of AI products (generated text, classification, next action, etc.) is probabilistic, meaning that when these products work, they are often 10x differentiators and when they don’t work, they are often duds, things don’t converge to the middle.
An example of a 10x differentiator: Earlier in 2024, Klarna conducted an analysis of its AI-driven customer service operations, and the results were impressive. By integrating AI-powered agents, Klarna was able to handle a significant portion of their customer interactions with minimal human intervention. Specifically, the AI system managed 2.3 million conversations, representing two-thirds of all incoming customer service chats.
This AI system did the work of approximately 700 full-time human agents, resulting in an estimated $40 million USD in profit improvements for Klarna in 2024.
At the same time, the cost of developing these models is substantial. Whether it's the computing power required to train a model or the infrastructure to support ongoing operations, costs remain high and do not diminish over time as they might in traditional SaaS models.
Let’s talk about what it really takes to build GenAI models. Training a model like GPT-4o or Llama 3.1 can cost anywhere from $100 million to $1 billion.
This means that some AI products might deliver outsized value, but others will be total duds, both requiring similar capital outlays – almost like blockbuster movie production. This uneven distribution of value creates a risky environment for companies building AI products.
All the while the economics are continually evolving. Two years ago, you might have been paying around $50 for every million tokens processed by a model like GPT-4o. Now, that cost has fallen to about $0.50 per million tokens. That’s a 100-fold reduction in cost. Open source models have seen a similar reduction in price and are much closer now in performance to models like GPT-4o. This indicates that it is likely that in some time the “blockbuster” dynamic in AI products will even out both in terms of output and cost overlays. Per my favorite VC, Tomasz Tunguz, highlights the dramatic reduction in AI costs, estimating that a sophisticated AI assistant, akin to the one in the movie Her, could cost about 78 cents per month in inference costs or $7 per month commercially—a fraction of a Netflix subscription.
But given current circumstances, just how do we go about pricing GenAI products?
For this we introduce a 5-step framework for pricing GenAI software products.
Effective pricing transformation begins with Step 1: Goals & Segments, which emphasizes the importance of aligning company objectives—such as boosting margins or capturing market share—and clearly identifying customer segments through Ideal Customer Profiles (ICPs). Misalignment within leadership on goals or ICPs can result in significant downstream challenges, including packaging and pricing strategies that fail to resonate with target customers. Achieving alignment ensures smoother operations and strategic clarity.
In Step 2: Positioning & Packaging, the focus shifts to designing offers that resonate with identified segments. Packaging must be based on a deep understanding of customer needs, avoiding pitfalls like "shelfware" (unused features) or over-granular offerings that slow sales cycles. Strategic packaging should prioritize customer value rather than quantity metrics.
AI products are often introduced as features within existing packaging lineups. For this it is useful to consider the rubric below (Jan Pasternak and I use a version of this in our course).
Step 3: Price Metric The next critical step is determining the pricing metric. This decision will shape how the AI product generates revenue and aligns with the broader business model.
There are different pricing options to consider. On the fixed side of the spectrum, as you see in the below Fig 3, traditional models like on-premises licenses or named user licenses offer predictable and stable pricing. These options are common in many SaaS models because they provide consistency and are easy to manage.
As you move toward more flexible pricing, consumption-based models start to appear. Companies like Amazon, for example, charge based on actual usage—whether it's for services like EC2 or S3. Similarly, OpenAI uses a token-based model where pricing varies depending on how much the system is used. These pricing structures offer flexibility but come with less predictability.
Choosing the right pricing metric is an art, but it is also further complicated by the fact that AI products have high running costs. This can make traditional user based pricing models a much harder selection because high use single users can drive up your costs enough for it to not make sense for you economically.
However, choosing the wrong usage/consumption based metric can also be problematic. Let’s say you are using my AI software. Whether you use it 60 times or 600 times across a few days, may not have a tangible impact on your success. However a 10x usage based invoice when you have not necessarily succeeded, is not going to make you a happy customer.
So, what here truly matters is selecting a pricing metrics that benefits both the parties:
Zendesk has recently started offering AI-powered agents that handle automated resolutions. In each plan, they provide a certain number of automated resolutions per agent each month. Beyond that limit, Zendesk charges up to $2 per additional automated resolution per month, or it can go as low as $1 if you buy a bundle.
Now, Zendesk has chosen "Automated Resolution per agent per month" as the metric for pricing. This is a foray in what is called “outcome based pricing”. This metric can be tricky because it may not have a consistent definition across their customer base. Zendesk might define it one way, but customers may have their own understanding of what counts as a resolution. It will likely take time for both Zendesk and its customers to fully agree on what this metric means. Resolutions are also unpredictable, and some can be very complex, while others can be easy, this affects the cost dynamics. More complex service environments could actually have Zendesk lose money at those customers.
How would you have approached the Zendesk case if it were up to you? It would necessarily involve creating pricing metric candidate options and weighing them against each other.
It seems Zendesk has chosen the most aligned metric, even though it comes with a lot of customer evangelism, instrumentation challenges and cost unpredictability.
Like in all things pricing, there is no wrong or right. Sometimes you just have to take a bet on something and be willing to listen to the market.
Step 4: Price Point Setting determines optimal price points, considering market conditions, customer willingness to pay, and competition, supported by testing before broad launch.
What is important to note here is that AI products are changing the benchmark price points established for traditional SaaS products. Tomasz Tunguz examines AI pricing strategies among SaaS companies offering copilots, revealing significant variance in AI add-on pricing relative to base product prices.
For instance, Google charges more for AI features than the base seat, while Loom adds about a 33% premium. He notes that companies like Microsoft and ServiceNow report AI features boosting productivity by approximately 50%, suggesting that to maintain revenue per customer, prices would need to double if headcount were reduced accordingly. This indicates that current pricing reflects an anticipated 40% productivity gain.
The premium charged by some of the add-on AI products is a positive step as it unshackles previously anchored price points e.g. $150/165 fully loaded per seat pricing for products like Zendesk or Salesforce.
Yet the cost complication inherent in price point setting is not gone. The table below shows the cost per call across different closed source and open source models.
The cost of AI-driven customer service tools varies significantly, especially as client numbers scale, with models like GPT-4o offering high-quality output but at a steep price. For instance, a hypothetical B2C company with 10,000 clients, each generating 3.65 million customer service calls annually, would face annual costs of $550 million if using GPT-4o, due to its high per-call processing costs of $0.0145. In contrast, leveraging an open-source model like Llama 3.1 BB would require an initial investment of approximately $377,000 for customization but stabilize annual costs at $12.36 million, as its per-call cost is only $0.00032. These examples highlight that running AI models at scale costs hard money and that technology design choices as of today are non-trivial. With GPT-4o, the inference cost is $55,000 for one client. Open-source models have a much lower inference cost—around $1,200 per year for one client.
Both price point setting and pricing metric selection will therefore have to remain very cognizant of costs and not just focus on the "willingness to pay” as is customary for most application layer SaaS products where the cost of hosting and running the software really does not scale the same way it does for GenAI products.
Finally, Step 5: Operationalization ensures pricing strategies are effectively implemented through systems like CPQs, ERP integration, and sales enablement. This step solidifies pricing success by aligning processes and teams, making it a critical component for long-term impact.
For this step the complications in a GenAI business are not much more complex than a regular SaaS business, the only difference being that they are much more likely to use some sort of usage metering. This requires product instrumentation, a suitable CPQ and billing systems, and an ability to be exact in what you charge – and not get the invoices incorrect when you bill customers.
As you can now see, technology choices impact the commercialization of a GenAI product and consequently how you package and price it.
A poor tech decision could seriously impact the cost of service, resulting in pricing models that fail in the market. And a good or bad decision is entirely context dependent. Closed source models might be great to prototype but for financial viability may need to be cutover to open source models as you scale.
Pricing metric decisions themselves are bringing both tech companies and customers to a brave new world where their prior mental anchors are going to be re-examined. It is indeed a fun time to be in tech.
To succeed with GenAI, it's essential to grasp the economics and pricing involved. The choices you make about which model to use, how you manage costs, and how you set your prices will make or break your product. No pressure!