Pricing design is just as important as product design when it comes to defining what your company actually does. Why? Because it forces you to clarify exactly what your customers are paying for. It forces you to begin at the end.
A healthcare IT firm signs a contract with a hospital promising to reduce its patient re-admission rate by a certain percentage. A space transportation company signs an agreement with NASA guaranteeing a particular cabin environment to transport lab rats to the ISS. A home security company promises its customers that they will never come home to a dark house.
We’ve been talking about outcome-based services in the Subscription Economy for a long time (those are all examples from my book). Start with a desired outcome, then organize and apply your people and technology accordingly. Service level agreements are replacing bills of sale. People want the milk, not the cow.
But now something interesting is happening – suddenly lots of software companies are taking advantage of agentic AI to create autonomous services that can analyze, respond, and take direct action in order to achieve a desired result. But many of these companies are not used to thinking in terms of actual results. Their invoices are still stamped “For Services Rendered.”
Think of a construction project. The site manager is responsible for making sure the project is completed safely, on time, and within budget. That site manager works with dozens of subcontractors — electricians, plumbers, HVAC suppliers, equipment rental companies — to make it happen. But the goal — the outcome — is a nice new building.
Suddenly, lots of subcontractors are talking like site managers.
Gartner predicts that 40% of agentic AI projects will be scrapped by 2027 (personally, I think that number is very low for such a nascent technology). “Most Agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype and are often misapplied,” said Anushree Verma, senior director analyst at Gartner. The company warns against “agent washing,” where vendors rebrand traditional tools as agentic AI without meaningful improvements. They suggest that only a few hundred out of thousands of new agentic AI services are actually legitimate.
Want to figure out which agentic services are real, and which are hype? I suggest you follow the money. Jump straight to their pricing page.
As my colleague Michael Mansard notes, there are several ways to price these services — per action or completed workflow (the consumption model), per agent (similar to headcount), or per actual result: “Is your AI more like a worker whose time should be billed? A service producing outputs that can be itemized? A utility consumed by the action? Or a partner delivering business outcomes who could be paid for success?”
I would argue that if you’re either a) building a new agentic AI service or b) evaluating a potential new one, you start with the outcome model. Today, many established companies with extensive resources, such as ServiceNow, Salesforce, Zendesk, and Intercom, charge for achieving specific business objectives. They directly align their pricing with specific value delivered.
That’s telling. Because it’s very hard to do. You need to be extremely confident in your capabilities to deliver a truly outcome-based service.
It’s fine to charge for a specific service rendered. In fact, lots of construction subcontractors talk about outcomes in their BPO contracts — energy efficiency, hot and cold running water, stable and secure communications, etc. We work with hundreds of companies that are taking advantage of agentic AI to reinvent themselves as solution providers.
But a lot of these new agentic AI services are essentially just dressed-up equipment rental companies. My advice is to head skip right over all their glossy product demos and head to their pricing page, to find out what they’re really offering.
Begin at the end.
