Price the Job, Not the Tool
A First Principles guide to deconstructing value and building outcome-based contracts for an AI-powered world.
Part 1 of a 6-Part Series on pricing based on outcomes
Introduction: The Ticking Time Bomb in Our Software
The dashboards glow in the conference room at 7:00 p.m. A Chief Financial Officer, a Chief Marketing Officer, and a VP of Operations are in their third hour of a tense budget review. On one screen is the consolidated expense report. The number is clear, predictable, and climbing: $50,000 per month for the new, enterprise-grade, “AI-Powered Customer Intelligence Platform.” On the other screen are the metrics that truly matter: customer churn rate (stagnant), qualified lead conversion (anemic), and net promoter score (flat).
The CMO breaks the silence. “The vendor says our team’s adoption is at 90%. They’re using the tool every day. We’re getting exactly what we paid for.”
The CFO looks at the two screens—the $50,000 cost and the flat-lining value—and replies, “Then we are paying for the wrong thing.”
This scene, playing out in boardrooms across the globe, is the quiet, ticking time bomb at the heart of the modern software industry. It is the sound of a paradigm collapsing under its own weight. We are in the grip of a profound and unsustainable contradiction: we are deploying 2030s technology—artificial intelligence—but we are paying for it with 1990s logic.
That logic is the seat license. It is the monthly subscription. It is the all-you-can-eat SaaS model that defined the cloud revolution. This model was a brilliant, disruptive innovation. It democratized access to unfathomably complex tools, moving software from a massive, one-time capital expenditure (think installing an Oracle database from a box of CDs) to a simple, predictable operating expense (think opening a Salesforce tab in your browser). For two decades, this model has served us well. It has built empires. And it is now becoming obsolete.
The problem is that the subscription model, by its very nature, prices access. It charges you for the right to use a tool, regardless of whether you use it well, or at all. The entire economic incentive of a traditional SaaS vendor is to sell more seats and prevent churn. Their success is measured by adoption, engagement metrics, and renewal rates. But as the CFO in our conference room knows, adoption is not a business outcome. Engagement is not profit. A 90% login rate is not a 10% reduction in customer churn.
This misalignment, this gap between what we pay for (access) and what we want (value), was a tolerable friction in the age of passive software. When software was just a tool—a glorified hammer or a smarter spreadsheet—it was still a human who was ultimately responsible for the outcome. The software assisted; the human did. You paid for the hammer, and you paid a human to swing it. The value was created by the human, and the software was a simple cost of doing business.
Artificial intelligence, particularly the generative and autonomous systems now coming online, shatters this model completely.
This new technology is not a better hammer. It is a carpenter. It is not a smarter spreadsheet. It is an autonomous analyst. It is not a passive tool that assists you in writing a marketing email; it is an active agent that you hire to run the entire email campaign. It is a system you don’t just use, but deploy. You don’t give it tasks; you give it objectives.
And you do not pay a carpenter a flat monthly fee for access to their toolbox. You pay them for the finished deck. You do not pay an analyst a subscription to use their brain; you pay for the actionable insight that saves your company $10 million.
This is the core of the tectonic shift we now face. The traditional pricing models of the last thirty years are fundamentally and irrevocably broken, because they are built to charge for access to a tool, not the value created by an agent.
This article is an argument for the inevitable, logical, and necessary successor to the SaaS subscription: outcome-based pricing.
We will argue that this is not simply a new pricing “tactic” or a clever “option” for your go-to-market strategy. It is the only model that realigns the incentives of AI providers and their customers. It is the only way to build sustainable, long-term partnerships in an economy where value creation is becoming automated. It is a model that changes the entire nature of the vendor-customer relationship, moving it from a transactional sale to a joint venture in risk and reward.
To get there, we must first deconstruct the old models and understand precisely why they fail. We must then build a new philosophy of value, grounding our work in the durable, powerful framework of “Jobs-to-be-Done,” which gives us the language to price the job, not the tool. We will provide a practical toolkit for deconstructing any business problem to find the true outcome to price against. And we will confront, head-on, the hard problems this new world creates: the challenges of measurement, attribution, alignment, and trust.
The journey from “price per seat” to “price per outcome” is not easy. It will require companies to build new technical infrastructures, new financial models, and new muscles for sales and negotiation. It requires vendors to have terrifying, unshakeable confidence in their product’s ability to deliver. And it requires customers to become true partners, willing to share in both the risk and the massive upside.
For the company in our dimly lit conference room, the path forward is unclear, but the problem is not. They are trapped, paying for activity while praying for results. The future will belong to those who have the courage to close that gap—to stop buying software and start hiring outcomes. The future will be built, and priced, by those who are willing to bet on the one thing that has ever really mattered: the result.
Part 1
Chapter 1: The SaaS Mirage: When “All-You-Can-Eat” Breeds Indifference
To understand why the subscription model is failing, we must first appreciate why it triumphed. It’s easy to forget that “Software as a Service” was a radical, revolutionary idea. It solved a series of acute, expensive problems that defined the on-premise era of the 1990s and early 2000s.
In that world, buying enterprise software was an act of extreme financial and operational commitment. It meant a massive, multi-million dollar, up-front capital expenditure. It meant buying racks of servers to host it. It meant a brutal, 18-month implementation cycle, managed by legions of high-priced consultants. And at the end of it all, you were stuck with that version—a static, depreciating asset—until the next painful, expensive upgrade cycle.
The SaaS model, pioneered by companies like Salesforce, changed everything. It was a masterstroke of access democratization. With nothing more than a credit card and a web browser, a 10-person startup could suddenly access the same sophisticated CRM tools as a Fortune 500 giant. The SaaS model transformed software from a product you buy into a utility you subscribe to.
Its psychological genius lay in its predictability. For the CFO, the unpredictable, spiky capital expenditure was smoothed into a manageable, predictable operating expense. For the vendor, the lumpy, hit-driven revenue of license sales was transformed into the holy grail of Annual Recurring Revenue (ARR). This financial innovation created the entire “cloud” economy, birthing a generation of unicorns and fundamentally reshaping our relationship with technology.
For two decades, this model worked beautifully. And it did so because of a tacit, universally accepted agreement: the vendor was responsible for the tool, and the customer was responsible for the outcome.
The model was perfect for a world of passive software. A subscription to Microsoft Office 365 gives you access to the workbench (Excel, Word, PowerPoint). It is still your analyst’s job to build the financial model that finds the $20 million error. A subscription to Adobe Creative Cloud gives you access to the tools (Photoshop, Premiere Pro). It is still your designer’s job to create the campaign that wins the market. The software is a passive enabler; the human is the active agent of value.
The subscription price is a fair, rational rent for access to a high-quality workbench. But this elegant model begins to warp and crack when the vendor’s incentives are examined more closely.
In a subscription model, the vendor’s primary financial goal is not to create value for the customer; it is to defend the renewal. The entire business, from its R&D priorities to its “Customer Success” organization, is built around this single objective. As long as the customer renews their subscription, the vendor is successful. Whether the customer actually extracted any value is a secondary concern, correlated but not causal.
This single, simple incentive gives rise to two pathologies that now define mature SaaS: feature-stacking and engagement-hacking.
Feature-stacking is the vendor’s primary lever for justifying the renewal. The R&D roadmap is driven by the need to add more. More buttons, more modules, more dashboards, more “AI-powered” widgets. The product is sold on the promise of its potential, not the proof of its performance. The goal is to make the product look so comprehensive that the customer believes it must be valuable. The product inevitably bloats, transforming from a sharp tool that solves a specific problem into a sprawling “platform” that does a thousand things poorly.
When the customer’s CFO asks the department head if they can cut the $50,000 monthly fee, the answer is mired in fear and ambiguity: “We can’t. The new ‘Synergy Dashboard’ is where we track our new KPIs. And the ‘Insights’ module is supposed to be critical for Q3. We haven’t used it yet, but we can’t risk losing it.” The subscription is renewed, not because it delivered $50,000 in value, but because its perceived complexity has made it “too sticky to cancel.”
This is where the second pathology, engagement-hacking, comes in. If a vendor cannot definitively prove value, they will settle for the next best thing: proving usage. This is the entire function of the modern “Customer Success” team. They are not, as the name implies, primarily focused on making the customer successful. They are focused on generating reports that show engagement.
They track logins, clicks, “time in-app,” and adoption rates. They congratulate the customer on achieving “90% team adoption.” They have successfully confused activity with achievement. The product itself is engineered for this. It sends a flurry of notifications, daily digest emails, and “insights” that require the user to log in. It is designed to be “sticky” in the same way a social media app is, hacking the user’s attention to create a data trail of “engagement” that can be presented at the renewal meeting. The user is busy, the dashboard is active, and everyone feels productive. But the company’s core metrics—churn, revenue, profit—remain untouched.
This is the great SaaS mirage. It’s an “all-you-can-eat” buffet that incentivizes the restaurant to pile the steam trays with cheap, starchy fillers. The customer is full, but not nourished.
This model was perfectly acceptable, even ideal, for content and passive tools. For Netflix, the “job” is access to the library. The subscription is the value. For a basic CRM, the “job” is to be a stable, available database. The subscription is a fair price for that reliability.
But artificial intelligence is not a passive database. It is not a library of content. It is an agent.
When you buy an “AI sales tool,” you are not hiring it to “be available” or “be sticky.” You are hiring it to find leads. When you buy an “AI logistics platform,” you are not paying for “access to the routing algorithm.” You are paying for it to cut your fuel costs.
This is the fundamental disconnect. The vendor is selling access to an “AI-powered tool,” and their incentives are to stack it with features and prove your team is “engaging” with it. The customer, meanwhile, is hiring an agent to do a job, and they only win if the job gets done.
In this new world, the subscription model is a con. It’s a misalignment of interests, a shell game that rewards the appearance of value over the delivery of it. It places 100% of the burden of value extraction on the customer and 100% of the risk. The vendor gets their $50,000 a month whether the AI generates $5 million in new business or sits in a browser tab, idly racking up “engagement” minutes.
The industry’s first attempt to solve this, usage-based pricing, seems like a logical next step. If paying for access is bad, surely paying for activity is better. But as we will see, this is just a more granular version of the same trap. It moves us from paying for the toolbox to paying for every swing of the hammer, all while the house remains unbuilt.
Chapter 2: The Usage Trap: Paying for the Hammer, Not the House
The subscription model’s flaws are obvious enough that a seemingly more rational alternative has already gained significant traction: usage-based pricing. This model feels fairer, more modern, and more aligned. The logic is simple: instead of paying a flat fee for access, you pay for what you actually use. This is the metered model of the cloud infrastructure world (think Amazon Web Services) applied to the AI layer. You pay per API call, per thousand tokens processed, per minute of compute, or per gigabyte of data stored.
On the surface, this solves the “all-you-can-eat” problem. The customer is no longer paying for a buffet of features they don’t use. The vendor can’t just coast on a renewal; they must deliver a service that customers find useful enough to activate. This model ties revenue directly to consumption, which, in theory, is a proxy for value. If customers are using the service a lot, they must be getting a lot of value from it. Right?
This is the great, seductive lie of usage-based pricing. It is a more granular, more precise, and more dangerous version of the very same trap. It has not solved the core misalignment. It has simply shifted the misalignment from access to activity.
The customer still does not want activity. They want an achievement. They do not want to run an API call; they want to get a correct answer. They do not want to process tokens; they want to generate a lead. They do not want to use compute; they want to solve a problem.
Usage-based pricing mistakes the cost of the labor for the value of the result. You are now paying for every swing of the hammer, not for the finished house. And in the world of AI, this creates a host of perverse, counter-productive incentives that punish efficiency, reward waste, and leave the customer holding all the risk.
Consider the incentives of an AI vendor who charges per thousand tokens. Their financial incentive is to maximize the number of tokens processed. Is it any surprise, then, that so many models are notoriously verbose? A user asks a simple question, and the AI responds with a five-paragraph preamble, a long-winded answer, and a two-paragraph summary. This feels comprehensive, but it’s also an act of economic self-interest. A more efficient, concise, and direct answer would be more valuable to the user, but it would be less profitable for the vendor.
This model actively punishes efficiency. Imagine you are a vendor with two AI models. Model A is a massive, 100-billion-parameter beast. It’s slow, expensive to run, and verbose. It takes 1,000 tokens to deliver a correct answer. Model B is a new, highly optimized 10-billion-parameter model. It’s fast, cheap, and brilliant. It delivers a better answer in just 100 tokens.
In a usage-based world, the vendor is financially penalized for deploying the superior model. Model A, the inferior product, is 10 times more profitable. The vendor is now in a direct conflict of interest with its customer. The customer wants the fastest, cheapest, and best answer. The vendor is incentivized to sell them the slowest, most expensive, and most long-winded one.
Now, let’s look at it from the customer’s side. The usage-based model transfers 100% of the performance risk to the customer. When you hire an AI agent to complete a task—say, “Analyze these 10,000 customer reviews and identify the top five reasons for churn”—you are now on the clock. You are paying for every token it processes, every API call it makes, every second of compute it burns.
What if the AI misunderstands the request? What if it hallucinates, gets stuck in a loop, or produces a shallow, useless analysis? The customer pays all the same. They pay for the effort, not the outcome. They are left with a massive bill and a problem that is still unsolved. The CFO from our introductory chapter is in an even worse position: her $50,000 monthly subscription was predictably expensive but useless. Her new usage-based bill is unpredictably expensive and still useless. She has swapped a known liability for a variable one.
This is the “taxi meter” problem. When you get in a cab, you are paying for miles and minutes—proxies for the “job” of getting from Point A to Point B. If the driver gets lost, hits traffic, or takes the “scenic route,” you pay for their inefficiency. You pay for the activity, not the achievement. The driver has no incentive to find the magic shortcut that gets you there in half the time, as it would cut their fare in half.
AI, in its current form, is the ultimate inefficient taxi driver. It’s a stochastic, non-deterministic system. You can run the same prompt five times and get five different answers, five different token counts, and five different costs. A usage-based model forces the customer to become a gambler, rolling the dice every time they hit “enter,” hoping for a good result but paying for any result.
This model is simply a form of cost-plus pricing disguised as innovation. The vendor calculates their underlying cloud compute cost (from AWS, Google, or Microsoft), adds a healthy margin, and bills the customer for it. They have successfully arbitraged their cloud bill, passing the compute cost—and all the risk of inefficiency—directly to their customer. They have created a business with no performance risk, no skin in the game. They are selling AI activity as a raw commodity.
This is not a partnership. It is a tollbooth.
The usage-based model is a half-step, a transition fossil. It correctly identifies that access is the wrong metric, but it incorrectly identifies activity as the right one. It’s a mechanic’s solution, focused on the inputs (compute, tokens, data) rather than the outputs (value, profit, solutions).
This is why, for high-stakes business problems, this model is ultimately a dead end. No rational executive will bet their company’s P&L on a system with a variable cost and a variable outcome. You cannot build a business strategy on a taxi meter.
To find a sustainable model, we must stop pricing the inputs altogether. We must stop paying for access to the toolbox and stop paying for the swings of the hammer. We must, for the first time, build a commercial model based entirely on the finished house.
To do this, we need a new language. We need a way to describe, measure, and agree upon the “finished house” before we build it. We must move our entire commercial vocabulary away from the vendor’s solution and toward the customer’s problem. This new language, this new philosophy of value, is called “Jobs-to-be-Done.”
I make content like this for a reason. It’s not just to predict the future; it’s to show you how to think about it from first principles. The concepts in this blueprint are hypotheses—powerful starting points. But in the real world, I work with my clients to de-risk this process, turning big ideas into capital-efficient investment decisions, every single time.
Follow me on 𝕏: https://x.com/mikeboysen
If you’re interested in inventing the future as opposed to fiddling around the edges, feel free to contact me. My availability is limited.
Mike Boysen - www.pjtbd.com
De-Risk Your Next Big Idea
Masterclass: Heavily Discounted $67
My Blog: https://jtbd.one
Book an appointment: https://pjtbd.com/book-mike
Join our community: https://pjtbd.com/join