When Seats Still Make Sense (and When They Don't)

There's a lot of talk recently about seat-based pricing being dead. As AI tools automate tasks once done by people, the idea is that charging per seat becomes as absurd as charging per desk in a remote-first company. Small teams with AI agents will generate enterprise-level output, and pricing models tied to headcount will collapse under the contradiction.

There's truth in this, but only partialy. The rush to abandon seats misses something important: seats were never arbitrary. For certain products, they captured real value. The question isn't whether to abandon seat-based pricing; it's understanding when seats reflect genuine value and when consumption tells a better story.

When Seats Create Value

Selling access to a team is fundamentally different from selling value to an individual user or AI Agent. For example, Slack delivers value very differently from chatGPT. Slack relies on mutliple cross collaboration between a team, the more users in that tool, the more value each person finds being able to collaborate. The value of a collaboration tool isn't just the features it offers each user. It's the interaction between all the members on the platform. This distinction matters for pricing.

Products with genuine network effects derive value from the connections between users, not just the capabilities available to each. A messaging platform is worthless with one user, marginal with five, and transformative with an entire organisation. The more people on the platform, the more valuable it becomes for everyone. Seat-based pricing captures this dynamic directly: as the group grows, so does the value, and so does the price.

Even without network effects, seat counts often correlate with value in ways that make pricing straightforward. The size of an organisation's team typically reflects its overall budget and the scale of problems it's solving. A project management tool used by three people serves a different scale of operation than one used by three hundred. The hundred-person company isn't just using more of the product; it's deriving more value from the coordination and visibility the tool provides.

Seat-based pricing performs several useful functions. It ties revenue to a value signal that customers can feel: they understand why adding users costs more because they can see the additional value those users receive. It enables viral, word-of-mouth growth as individual users bring the tool into their teams. And it gives organisations clear control over usage and cost through a single lever: the number of people with access.

When Consumption Creates Value

Tying value to revenue works differently when value comes from actions rather than access. Consumption-based pricing shines when high-cost features deliver value in proportion to how much they're used, and when that usage varies unpredictably.

The most visible example is AI. Every API call to a language model consumes compute resources and generates tokens. The cost structure is variable, and so is the value delivered. A customer making a thousand API calls derives different value than one making a million. Charging per token aligns the price with both the cost incurred and the value delivered.

Consumption billing tends to work when three conditions hold: the features being consumed are genuinely high-value, that consumption varies significantly between customers or over time, and users can control their consumption. If everyone consumes roughly the same amount, a subscription makes more sense. If users have no control over their usage, variable pricing feels punitive.

In B2B contexts, consumption pricing offers something enterprises crave: predictability through control. If a company can predict that using a service will drive a certain percentage of growth, they can scale usage and revenue together. The cost becomes a known input to a growth equation rather than a fixed overhead. This predictability extends to cost optimisation. When businesses have choices between services at different price points, like "thinking" models versus their faster, cheaper counterparts, they can match the tool to the task and manage their margins accordingly.

Why AI Changes the Equation

The introduction of AI features into SaaS products has made per-user costs variable in ways they never were before. A user who triggers a hundred AI-assisted actions costs dramatically more to serve than one who triggers ten. The seat that was once a reasonable proxy for value now conceals wildly different consumption patterns.

This creates pressure from multiple directions. Some features are becoming significantly more valuable with AI augmentation. A single button that summarises a document or drafts a response can save hours of work. Others are becoming commoditised as AI makes basic functionality trivial. The value landscape is shifting beneath existing pricing structures.

Customers notice this shift and want more control. In a seat-only model, the only lever available is adding or removing users. Changing the price per seat requires demonstrating increased value to every single user, a high bar when different users derive different value from AI-enhanced features. Customers want more levers to pull: the ability to dial up AI usage when it's valuable and dial it down when it's not, without the blunt instrument of removing access entirely.

The result is that seat-based pricing alone is no longer sufficient for products with significant AI components. It's not that seats have become meaningless; they haven't. It's that they can no longer carry the full weight of pricing when usage patterns vary so dramatically.

The Interface is changing

The way users interact with SaaS products is changing just as quickly as the cost structures behind them.

For years, SaaS has been built around the assumption of a human user logging into the tool. Pricing followed naturally from that model. A “user” was someone with a login, a seat, a defined place in the interface.

That assumption is breaking down.

Increasingly, products are being consumed programmatically. Developers are working directly from the terminal. Teams are interacting with tools through chat interfaces. Agents are orchestrating workflows across multiple services without anyone opening a browser. The interface is no longer the product. The outcome is.

In this world, the concept of a “seat” starts to feel artificial. You might have a single engineer wiring together APIs to generate reports, trigger workflows, or enrich data. That work could deliver value across an entire organisation, but it maps to exactly one user in a seat-based model. There is no natural moment where you would add more seats, because the value isn’t tied to more people logging in. It’s tied to more being done.

AI accelerates this shift. When agents can plan, execute, and iterate on tasks independently, they become the primary consumers of SaaS products. Humans move up a level, supervising and directing rather than directly interacting. The “user” becomes harder to define. Is it the person who set up the agent, or the agent itself?

Projects like x402 are early signals of where this is heading. We can explore a world where services are consumed directly by machines, with authentication, payment, and access handled programmatically. In that model, there is no concept of logging in, no dashboard, and no seat. Just capability, accessed on demand.

This creates a fundamental mismatch with seat-based pricing. If your most valuable customers are interacting via APIs, agents, or automated workflows, then pricing per human user becomes increasingly disconnected from how value is actually created and consumed.

Seats don’t disappear overnight, but they become less central. As interfaces fade and agents take over more of the interaction layer, pricing needs to follow usage, outcomes, or capabilities rather than the number of people who happen to have an account.

The Path Forward

The answer isn't abandoning seats for pure consumption pricing. That would mean giving up everything seats do well: the revenue predictability, the alignment with team growth, the familiar model that customers understand. The answer is combining models to match how your product actually delivers value.

Flexibility in your pricing model is now crucial. A collaboration tool might charge per seat for platform access (because the value of collaboration genuinely scales with team size) while adding consumption pricing for AI features that vary dramatically in usage. A data platform might combine a base subscription for access with metered pricing for compute and storage. The goal is matching each pricing component to the type of value it captures.

This isn't just about finding the right balance between seats and consumption. It's about recognising that different parts of your product may deliver value in fundamentally different ways, and pricing each accordingly. The platform itself might justify a flat fee. Team growth might justify per-seat charges. AI-powered features might justify usage-based metering. Combining these creates a pricing structure that reflects the actual value exchange rather than forcing all value into a single metric.

The competitive advantage goes to companies that find the right match, whether monetising the number of people, the activity they produce, or both. Rigid single-model pricing leaves value on the table. Flexible pricing that can adapt as products evolve and as customers' needs change captures more of the value being created.

Building Flexible Pricing Without Engineering Complexity

The challenge with hybrid pricing is usually the implementation. Tracking seats is straightforward. Metering consumption is more complex. Combining them in a single subscription, with proper proration and coherent invoicing, is substantial engineering work. Many companies default to simpler models because it's easier even when hybrid pricing would be more sensible or give the customer greater value.

Salable's Line Items solve this problem. Instead of building custom logic for each pricing component, you compose plans from building blocks: flat-rate for predictable charges, per-seat for team-based pricing, metered for consumption-based billing. A single subscription can combine a base platform fee, per-seat charges for team growth, and per-token pricing for AI usage, all managed through configuration rather than code.

When pricing is configuration rather than engineering, you can adapt as you learn. The hybrid structure that works at launch may need adjustment as AI capabilities evolve and customer usage patterns become clearer. Building that flexibility into your infrastructure means pricing can evolve with your product rather than constraining it.

Seats aren't obsolete. Consumption isn't the only answer. The products that thrive in the AI era will be those that match their pricing to how they actually create value and have the infrastructure to adapt as that value evolves.