The outcome-based pricing playbook: When it works, when it fails, and what to do instead

It wouldn’t be a stretch to say that outcome-based pricing is having a moment. Every AI pricing conversation eventually circles back to it. And why not? The promise is indeed seductive: Charge customers for results, not usage. This aligns your incentives perfectly with theirs, and you get to capture more value as your product gets better.

However, there’s just one problem: Almost nobody is actually doing it.

Only about 4.5% of SaaS companies have implemented outcome-based pricing at scale. In working with AI and SaaS finance leaders, here’s what we’ve figured out: there’s a significant gap between the sales pitch and the billing reality. Many companies talk about outcome-based pricing, but very few actually bill that way. What most end up doing is usage-based pricing dressed up in outcome-based language.

That’s not necessarily a bad thing, but it’s worth understanding the difference.

In this article, we’ll examine why outcome-based pricing fails for most companies, and what practical alternatives exist to give you a clearer sense of whether outcome-based pricing might be viable for your business, or whether you’re better off with something else entirely.

What is outcome-based pricing?

If you think outcome-based pricing is some bleeding-edge concept dreamed up by AI startups, think again. The model has been around for over 60 years.

Our CEO has traced the origins of outcome-based pricing back to an unlikely source: jet engines.

“In the 1960s, Rolls-Royce launched the first ‘power by the hour’ concept. Before that, airlines would buy engines outright. As you can guess, it was a massive CapEx investment. So, Rolls-Royce flipped the model; instead of selling hardware, they’d charge airlines for the time the engine was actually powering thrust. Simply put, ‘You fly, you pay’, and vice versa."  – Alok Goel, Drivetrain Co-founder & CEO

The power-by-hour model worked so well that it became the industry standard for aviation. It was successful because the outcome was defined as a single, indisputable metric. 

So why hasn’t SaaS been able to replicate it? The short answer is that software “outcomes” are messier than flight hours.

What outcome-based pricing means today

Outcome-based pricing, also called performance-based or success-based pricing, is a model in which customers pay based on results achieved rather than seats purchased, features accessed, or resources consumed.

The key distinction is that payment is tied directly to measurable business results, not inputs or activity—the keyword being “measurable.” 

A customer using your product heavily but getting poor results would pay less. A customer using it lightly but seeing great results would pay more. That’s a fundamentally different value exchange than traditional SaaS models.

What counts as an outcome depends entirely on the product and the problem it solves. For a customer support platform like Intercom’s Fin, that might be resolved tickets, a support conversation either got resolved or it didn’t. For a fintech product, it could be the number of fraud attempts prevented or transactions processed. For an AI SDR tool, it might be qualified leads generated or meetings booked. For an accounts receivable agent, it could be cash collected, and so on.

The common thread: the customer pays for what they got, not what they used. In theory, this creates perfect alignment; your success is literally their success. In practice, as we’ll explain, it’s rarely that clean.

How outcome-based pricing compares to seat, usage, and value-based models

To understand where outcome-based pricing fits, it helps to see it alongside the models it’s meant to replace (or complement, in some cases). The table below covers the three most common pricing models in SaaS today, alongside outcome-based pricing:

As you can see, each model makes different tradeoffs around risk, predictability, and alignment with customer value. Which model is best for your business really depends on the scenario. 

Here are a few important nuances to keep in mind:

• Seat-based pricing breaks down for AI because agents don’t log in, don’t need licenses, and can generate unlimited work once deployed, which disconnects price from both cost and value.
• Usage-based pricing maps cleanly to AI cost structures, but the units that matter to vendors, like tokens, calls, and compute, don’t really align with how customers perceive value.
• Value-based pricing works best as a sales and negotiation framework for AI, but it doesn’t work well as a billing mechanism because ROI is often delayed and hard to verify objectively.
• Outcome-based pricing only holds when outcomes are quantifiable, vendor-controlled, verifiable at scale, and when attribution is defensible. 

When outcome-based pricing actually works

According to Goel, true outcome-based pricing isn’t viable for most AI SaaS businesses.

“Maybe 5-10% of companies can actually pull it off. The rest will end up somewhere on the usage-based spectrum, even if their marketing says otherwise.”  – Alok Goel, Drivetrain Co-founder & CEO

So what separates the 5% from everyone else?

The four criteria for outcome-based pricing success

For outcome-based pricing to work, your product needs to clear all of the following four hurdles to legitimately say you are using a true outcome-based pricing model:

1. Outcomes must be clearly quantifiable: The outcome you’re pricing on has to be measurable with clear, agreed-upon metrics.
2. Attribution must be defensible: The outcome must be directly attributable to your product and your product only. 
3. The vendor must have meaningful control: Outcome-based pricing only makes sense if you have meaningful control over whether the outcome actually occurs.
4. Measurement must be scalable: The measurement infrastructure needs to be automatic, consistent, and defensible for every customer, every time.

A decision framework for CFOs 

Before you commit to outcome-based pricing, or any model with outcome-based components, run through the diagnostic questions in the table below. 

If you answer “no” to any of these questions, pure outcome-based pricing probably isn’t right for your business. That doesn’t mean you can’t use outcome metrics; it means you should use them for value demonstration, not billing.

‍

Modern examples of outcome-based pricing in AI SaaS

So, where is outcome-based pricing actually working? 

In each of the rare cases where outcome-based pricing works, the same four conditions are met at the same time: the outcome is precisely defined, attribution is defensible, the vendor controls the drivers of success, and measurement scales without exception.

Intercom’s Fin

Intercom’s AI customer service agent, Fin, is priced based on a simple outcome: Did the conversation get resolved, or didn’t it? It’s binary. Either the customer’s issue was handled without human intervention, or it wasn’t.

This works because resolution is:

• Measurable (a simple yes or no).
• Clearly attributable to Fin (the AI handles the whole conversation).
• Within Intercom’s control (the agent either resolves it or escalates).
• Easy to track at scale (every conversation has a disposition).

For a CFO making that buying decision, it’s super easy to value what a resolved call is worth.

The trade-off is that by keeping the outcome simple, you ignore other dimensions of success, such as customer satisfaction or sentiment. 

Simplicity is exactly what makes the model work.

Narrow AI agents

Outcome-based pricing also works well for AI agents that handle narrow, well-defined tasks. The kind of work you wouldn’t want to dedicate a full-time human to do.

Think accounts receivable automation (cash collected), data enrichment (records verified), or invoice processing (invoices matched and approved). These are discrete, countable outcomes with clear attribution. The agent did the work, or it didn’t. There’s no ambiguity about who deserves credit.

The narrower the task, the cleaner the outcome. 

Credit card processing

An often-ignored fact is that outcome-based pricing isn’t new to finance. Credit card processors have been doing it since the beginning.

When a merchant swipes a card, the processor takes a percentage of the transaction. The outcome, a completed payment, is perfectly measurable, directly attributable, fully within the processor’s infrastructure, and scales automatically. The merchant can’t hide transactions; every outcome is visible to both parties.

It’s such a clean model that we don’t even think of it as outcome-based pricing. We just call it payment processing.

The common thread is that all these examples are atomic, binary outcomes that are easy to measure and impossible to dispute: resolved or not, collected or not, processed or not.

The more dimensions you add, the more questions you have, and the harder outcome-based pricing becomes. Was the customer satisfied? Did the lead convert downstream? Did the output drive real business value? These examples work because they keep it simple on purpose.

How outcome-based pricing breaks down in practice

The examples above work because they’re exceptions. For most AI SaaS companies, outcome-based pricing begins to break down, not as an obvious failure, but with exceptions. These might include outcomes that technically occurred but don’t quite meet the agreed definition, customers who dispute attribution, or finance teams pulled into manual review to resolve edge cases that were never meant to be revisited.

As these exceptions begin to stack up, the model you thought was scalable isn’t anymore. Billing cycles start slowing down, and your forecasts become harder to trust. The sales team pushes for more flexibility to close deals, while finance absorbs growing variability and revenue unpredictability. 

Attribution and control challenges

The biggest killer of outcome-based pricing isn't technology or billing infrastructure. It's the fundamental difficulty of proving that your product caused the outcome.

The multi-dimensional value problem

Most SaaS products don’t operate in isolation. They deliver value through complex workflows involving multiple tools, human effort, and external factors. When an outcome occurs, it’s often impossible to attribute it cleanly to a single product or person.

Consider a sales enablement platform. Let’s say you close a deal. Was it the enablement tool that made the difference? The CRM? The rep’s skill and relationships? Market timing? Competitive dynamics? In reality, all of those factors probably had some impact on the result, which means none of them can credibly fully claim the outcome.

This problem intensifies as products become more sophisticated. For example, let’s say you have an AI SDR that handles research, enrichment, delivery, and conversion. Each of these capabilities produces an incremental outcome that contributes to the end result—the final outcome. 

So how do you price that? Do you bill based on the incremental outcomes? That becomes problematic if the sale falls through somewhere along the line and the customer doesn’t get the outcome they expected. Or, do you bet all your revenue on the final outcome? 

The thing is, a human BDR or SDR provides value in a multi-dimensional way rather than a simple binary, resolved or not resolved. This kind of value, delivered via AI, is difficult to price in a way that is both compelling and makes sense to the parties involved.

The gaming problem

Outcome-based pricing creates a visibility problem—Customers don’t have any incentive to give you full visibility into their outcomes. 

Goel’s experience at Google illustrates this well.

“About 18 years back, I used to work at Google in the ads department. Back then, my firm view was that, over time, the world would become more outcome-based. Of course, that didn’t happen. The advertising world stayed stubbornly usage-based: cost per click, not cost per acquisition.” – Alok Goel, Drivetrain Co-founder & CEO

He explained that Google wanted to charge based on cost per acquisition, and everyone wanted full visibility, full tracking, and full ROI for every marketing dollar spent. 

Why? Goel said it was partly because of the attribution problem. Everyone defined “outcome and “acquisition” in different ways. He added that there was also no real incentive for customers to share their data, “Advertisers wouldn’t give Google enough visibility into their actual results. They figured if Google knew exactly how much value advertisers were getting, they’d use that information to raise prices.”

The parallel with today’s AI SaaS industry is hard to miss. While not a true outcome-based model, the structural barriers that prevented Google from charging based on cost per acquisition—attribution complexity, definitional disagreements, and visibility issues—exist for the majority of SaaS companies today.  

So, if you can’t reliably bill on the outcome itself, the real question becomes: What metric captures the value your customers want and still scales cleanly?

The control problem

Outcome-based pricing only makes sense when you have meaningful control over outcomes. But what happens when success depends heavily on factors outside your influence?

This includes things like customer implementation quality, data hygiene, how their sales team follows up on leads, their product-market fit, seasonality in their business, and macroeconomic headwinds.

If you’re pricing on outcomes, you’re basically exposing your business to all of it. A customer with a weak sales process might generate terrible results from a product that works beautifully for everyone else. With outcome-based pricing, you’d be forced to eat that loss.

The margin erosion problem

AI costs money every time it runs. This is a financial reality that outcome-based pricing enthusiasts often gloss over.

Unlike traditional software with near-zero marginal cost, AI features incur real costs (API calls, inference, compute) with every interaction. If pricing is tied to outcomes rather than usage, vendors have no choice but to absorb these costs regardless of whether outcomes materialize. 

This creates a particularly nasty problem with power users. A small number of heavy users can drive disproportionate costs while producing average outcomes. Under usage-based pricing, that’s fine as they pay for what they consume. Under outcome-based pricing, you’re subsidizing their usage and hoping it pays off.

In either case, effectively managing AI infrastructure costs is crucial in these rapidly evolving times. 

Research from ORB shows that ~85% of SaaS offerings now layer in usage-based pricing specifically to avoid margin collapse. CFOs and investors scrutinize gross margins closely, and outcome-based models introduce a level of unpredictability that can alarm boards and affect valuations.

The risk management nightmare 

If you’re considering an outcome-based pricing model, keep in mind that the risk is all on you, whether your product works or not.

Think about what that means operationally. You’re essentially becoming an insurance company, underwriting the risk that your product delivers results. If you’re getting 100 customers, how many will actually succeed? 

Can you actually guarantee ROI in an objective way? For most products, the honest answer is no, as there are too many variables and a lot depends on customer factors such as their sales cycle, product quality, market, and execution.

The tension between what customers want and what actually scales

The fundamental dilemma every AI SaaS company faces is the fact that what customers want and what your business needs are often in direct conflict.

On one extreme, they want predictability. They are willing to pay “X” dollars per seat and have a clear understanding of how much they are paying. They can then evaluate whether the product works, but at least there are no surprises.

On the other extreme, they’re happy to pay for pure outcomes. You (vendor), for example, can give them infinite leads, and they’ll pay infinite money. 

Either of these is fine from the customer’s perspective. What they don’t want is to pay for your internal cost structure. They don’t care how many tokens you’re using, how much GPU time you’re consuming, or how efficiently you’ve architected your inference pipeline. That’s your problem, not theirs.

But, from the vendor’s perspective, your costs scale with usage, not with outcomes or seat counts.

Seat-based pricing satisfies buyer psychology but fails vendor economics in AI-heavy products. It’s like an all-you-can-eat buffet. Customers pay a fixed price, then consume as much as they want. For AI-powered products with real marginal costs, this model is a slow bleed. You cannot survive on seat-based pricing long-term in the agentic world.

Pure outcome-based pricing puts all the risk on you. And as we've covered, a huge chunk of software companies can’t objectively prove ROI. You’d be betting your revenue on outcomes you don’t fully control, while absorbing costs you definitely incur.

So, if this is the reality, what actually works?

Look for the Goldilocks metric—a pricing unit that scales with your cost structure, so you don’t get destroyed by heavy users, but feels like an outcome to customers, so they see the connection to value.

The question you need to ask yourself is: Can you find a metric that scales with your cost structure, is truly usage-based, and can be sold as an outcome-based model? Whoever gets that balance will actually succeed. Otherwise, you’ll end up flipping between usage and outcome-based.

The hybrid alternative: combining the best of both worlds

If pure outcome-based pricing is too risky and seat-based pricing doesn’t work for AI SaaS economics, what’s left?

The answer most successful AI SaaS companies are landing on is hybrid models that price on capacity but deliver value on outcomes.

Billing is based on inputs: capacity, credits, and usage, but the value story is all about outcomes. You’re not charging for meetings booked, but you’re absolutely showing customers how many meetings they’re getting. Similarly, you’re not billing per lead generated, but you are surfacing time saved, pipeline created, and ROI achieved at every opportunity.

It’s the most practical solution to what seems like an otherwise impossible tradeoff.

How to structure a hybrid pricing model

The most effective hybrid models combine three layers, each serving a different purpose.

1. Base subscription

A fixed platform fee for core access, support, and baseline functionality. This gives customers the budget predictability they crave and gives you stable revenue to cover fixed costs.

Think of it as the foundation. Customers know what they’re paying for access, regardless of how much they use.

2. Usage-based tier

‍Variable pricing tied to consumption of AI features: API calls, tokens processed, agent actions, and credits consumed. This is where you align your revenue with your costs.

Heavy users pay more, light users pay less. Your gross margin stays intact regardless of consumption patterns, and customers feel like they’re paying for what they actually use, which feels fair.

3. Outcome metrics layer

This is arguably the most important layer, where you track and report on business outcomes: ROI, time saved, revenue influenced, leads generated, etc., without tying billing directly to these metrics.

This layer proves the value of your product. It satisfies customer demand for transparency and ROI validation without exposing you to attribution disputes, gaming, or margin erosion.

The outcome metrics layer makes renewal conversations easy. It turns customer success check-ins into upsell opportunities and justifies your pricing. 

Using outcomes as a narrative, not a billing mechanism

This is what we call “outcome-based storytelling,” using outcome metrics to demonstrate value without making them the basis for billing. You capture the benefits of outcome alignment (customer trust, clear value demonstration) without the risks (attribution disputes, gaming, unpredictable revenue).

This is how it could work in practice:

• In sales conversations: Use outcome projections and case studies to justify your pricing. Show prospects what results look like, but when it’s time to sign, the contract is based on usage or subscription, something you can actually deliver and measure.
• In customer success: Build dashboards that show outcomes achieved: time saved, pipeline generated, and issues resolved. Customers see the value in concrete terms, which supports retention and expansion conversations, but the billing stays tied to usage.
• At renewal: Reference the documented outcomes when discussing contract extensions. Saying “You’ve generated 2,400 qualified leads through the platform this year” is a powerful argument for renewal, even if you’re billing by credits consumed, not leads delivered.

This might sound like marketing dressed up as a pricing model. And in a sense, it is. You’re selling the outcome story while billing on usage. This strategic separation is what makes this approach practical. 

“Customers want to see the value your product delivers, that’s non-negotiable. But they don’t need their invoice to be a math problem tied to outcomes they can’t predict. What they need is confidence that they’re getting ROI. You can give them that confidence through transparent outcome reporting without betting your entire revenue model on metrics neither side fully controls.” – Alok Goel, Drivetrain Co-founder & CEO

The billing mechanism and the value story don’t have to be the same thing. In fact, for most AI SaaS companies, they probably shouldn’t be.

Pragmatism over purity 

In the 1960s, Rolls-Royce invented “power by the hour” and transformed how airlines buy jet engines. Outcome-based pricing became the industry standard, and 60 years later, it still is.

In the 2000s, when Goel was convinced online advertising would evolve toward cost-per-acquisition. But the industry stayed usage-based.

So which storyline will AI SaaS follow?

Goel thinks it will be a bit of both, “If your product can truly deliver outcomes, you will kill the entire competitive landscape. You will end up being a monopoly in that market. You’ll be able to price based on outcomes.”

But that’s a big “if.”

"Maybe 5-10% of AI SaaS companies can pull it off, and those will be narrow, niche markets where outcomes are binary, attribution is airtight, and the vendor has real control." – Alok Goel, Drivetrain Co-founder & CEO

The other 90%, Goel said, will be better off with the hybrid approach we’ve described here, using outcome-based pricing as a narrative while billing on usage. They’ll track outcomes religiously, surface ROI constantly, and build their entire value story around results delivered. 

But when the invoice goes out, it’ll be based on credits consumed, not leads generated.

And that’s a pragmatic solution. 

The companies that win will be the ones who find a usage metric that aligns with their cost structure, feels fair to customers, and supports a compelling outcome story. They’ll price on what scales and sell on what matters.

CFOs need to guard against getting pulled in by the outcome-based pricing hype. CFOs are the ones who must ask the hard questions, run the diagnostics, and run through all the scenarios. If you can’t clear all the hurdles (which most companies can’t), then build a hybrid model that protects your margins while proving your value.

Managing the complexity of hybrid pricing models: tracking usage metrics, surfacing outcome data, and telling a coherent value story requires financial infrastructure that can handle it all. 

We can help you do that. With Drivetrain , you can bring your usage data, cost structures, and outcome metrics together in one platform, making it easy to model different pricing scenarios and demonstrate ROI to customers.

Check out Drivetrain to learn more about how we're helping SaaS businesses achieve sustainable growth in the AI Era.