The Seat Was Always a Proxy
Ross Sylvester, Co-Founder & CEO, Adrata | Feb 2026 | ~10 min read
In 1999, Salesforce made a simple pricing decision that defined an industry: $65 per user per month. It was elegant. It was legible. And it contained an assumption that nobody questioned for twenty-five years: that the value a piece of software delivers scales linearly with the number of humans who use it.
That assumption was always wrong. We just didn't have a better way to price.
The Proxy Problem
Per-seat pricing is a proxy. It approximates the value a customer receives by counting the number of people who log in. The logic feels intuitive: more users means more value extracted, which justifies a higher price. But the logic breaks down the moment you examine it.
Consider two companies, both paying $150/seat/month for a revenue intelligence platform:
Company A has 50 seats. Their reps use the platform to look up contact information before calls. Average session: 4 minutes. The software is a search engine with a CRM wrapper.
Company B has 50 seats. Their reps use the platform to identify hidden stakeholders in buying committees, map decision-driver profiles, and route deals through multi-threaded engagement sequences. Average session: 22 minutes. The software reshapes how they sell.
Same price. Radically different value. The seat count is identical, but the economic impact differs by an order of magnitude. Company B might generate $3M in incremental revenue from the platform. Company A might generate $200K. Both pay $90,000 per year.
This is what happens when you price on a proxy instead of on value. The proxy treats all usage as equivalent. It cannot distinguish between a login that produces nothing and a login that closes a $500K deal.
Why We Tolerated It
Per-seat pricing persisted because the alternative was worse. Before AI, measuring the actual value delivered by software was prohibitively difficult. You could track logins, page views, and feature adoption, but you couldn't attribute revenue outcomes to specific platform interactions with any precision.
The seat was a reasonable compromise. It scaled with organization size, which loosely correlated with value. It was easy to budget. It made contracts predictable. And because every competitor priced the same way, customers had a common unit of comparison.
But "easy to budget" is not the same as "correctly priced." The entire SaaS industry optimized for legibility at the expense of accuracy.
The consequences were structural:
-
Vendors under-captured value from power users. The rep who uses buyer group intelligence to close a $1M deal pays the same as the rep who checks dashboards once a week.
-
Vendors over-charged light users. Companies bought seats for people who logged in twice a month, inflating churn and depressing NPS.
-
Expansion was artificial. Revenue growth came from adding seats---adding humans---rather than from delivering more value per interaction.
-
AI adoption was punished. When a platform with AI agents can do the work of five humans, per-seat pricing rewards customers for not adopting the most valuable capability. You reduce seats to reduce cost, which means the vendor earns less precisely when they deliver more.
What AI Changes
AI does not incrementally improve the per-seat model. It breaks the model's foundational assumption.
The assumption was: value scales with humans. AI inverts this. Value scales with intelligence applied to decisions, and that intelligence can be delivered to one person or fifty, to a human or an agent, to a single critical moment or a continuous background process.
Three structural shifts:
1. The Unit of Value Becomes the Outcome
When AI can identify the hidden economic buyer in a deal, the value isn't "one seat used the platform." The value is "one deal saved from a no-decision because the right person was engaged at the right time."
This means pricing can---and should---connect to what the customer actually receives: deals influenced, stakeholders identified, pipeline accelerated, revenue protected. Not headcount.
The technology to measure this now exists. AI systems can attribute specific revenue outcomes to specific platform interactions with reasonable precision. The proxy is no longer necessary.
2. Flexibility Replaces Uniformity
Per-seat pricing is inherently uniform. Every seat costs the same regardless of what it's used for. This made sense when software did roughly the same thing for every user.
AI-powered platforms don't work that way. The intelligence layer adapts to each user's context:
- A CRO might use buyer group intelligence to audit the entire pipeline for authority gaps. The value is strategic and organization-wide.
- An AE might use it to map a specific buying committee in a $400K deal. The value is tactical and deal-specific.
- An SDR might use it to identify the right entry point at a target account. The value is operational and high-volume.
The right price for each of these interactions is different---not because the features differ, but because the outcome value differs. AI makes it possible to deliver precisely what each user needs, which means pricing should reflect that precision.
This is what flexible pricing actually means. Not "discount tiers" or "good-better-best packaging." It means the price reflects the specific value delivered in the specific context. The platform knows what it produced. The pricing should too.
3. Agents Don't Have Seats
This is the most straightforward problem and the one that will force the transition fastest.
An AI agent that monitors your pipeline for at-risk deals, identifies missing stakeholders, and generates engagement recommendations does not log in. It does not have a seat. It runs continuously, processes thousands of signals per hour, and delivers intelligence to whoever needs it.
How do you price that per-seat?
You don't. You price it per deal influenced, per insight delivered, per dollar of pipeline protected. Or you price it as a platform fee with usage-based components tied to the volume and complexity of intelligence consumed.
The companies that figure this out first will have a structural advantage. They can invest more in AI capabilities because their revenue model rewards the investment. Companies stuck on per-seat pricing will face the inverse: every AI improvement reduces their addressable seat count.
The Architecture of Flexible Pricing
The transition from per-seat to value-based pricing is not a packaging exercise. It requires infrastructure.
Measurement. You need attribution. Which platform interactions led to which outcomes? This requires connecting engagement data (who used what, when, how) to revenue data (which deals closed, which expanded, which were saved from stalling). The same buyer group intelligence that maps stakeholder engagement can map value delivery.
Granularity. You need to define the units of value for your product. For a revenue intelligence platform, these might be:
| Unit of Value | What It Measures |
|---|---|
| Stakeholders identified | Buyer group members discovered and enriched |
| Deals influenced | Deals where platform intelligence changed the approach |
| Authority gaps closed | Missing decision-makers identified and engaged |
| Pipeline protected | At-risk revenue flagged before it was lost |
| Time compressed | Days saved in sales cycle through precision targeting |
Transparency. Customers need to understand what they're paying for and why. This is where most usage-based pricing fails---it becomes opaque and unpredictable. The solution is to make the value visible. Show the customer exactly what the platform produced: "This quarter, we identified 47 stakeholders your team hadn't engaged, across 12 deals worth $4.2M in pipeline. Three of those deals closed that would have stalled without multi-threading."
When customers can see the value, they don't object to paying for it.
The Hybrid Model
Pure usage-based pricing is not the answer either. Customers need budgetary predictability. CFOs need to forecast costs. Procurement needs a number they can approve.
The emerging model is hybrid: a platform fee that provides baseline access plus value-based components that scale with outcomes. Think of it as a retainer plus performance.
The platform fee covers infrastructure: data access, CRM integration, baseline analytics, user accounts. This is predictable and budgetable.
The value-based component covers intelligence: buyer group identification, deal coaching, predictive analytics, agent-generated insights. This scales with the volume and complexity of intelligence consumed.
Bessemer's data shows that hybrid models (subscription + usage) deliver approximately 21% median revenue growth versus 13% for pure subscription. The companies adopting hybrid models are growing faster because their revenue architecture aligns with their value delivery.
What This Means for CROs
If you're a CRO evaluating AI-powered revenue tools, ask three questions:
First: Does the vendor's pricing model align with the value they claim to deliver? If they say "we help you close more deals" but price per-seat, there's a disconnect. The vendor is not confident enough in their own value to tie their revenue to it.
Second: Can the vendor measure the value they deliver? If they can show you---with data---which deals were influenced, which stakeholders were identified, which pipeline was protected, then value-based pricing is possible. If they can't, you're buying a promise, not a system.
Third: Does the pricing model punish AI adoption? If adding AI agents to your workflow reduces seats and therefore reduces the vendor's revenue, the vendor has a structural disincentive to make their AI better. You want vendors whose economics improve when their AI improves. That alignment only exists in value-based models.
The seat was a reasonable proxy for an era when we couldn't measure what software actually produced. That era is ending. AI doesn't just change what software can do---it changes how we know what software did. And once you can measure value, pricing it on a proxy is not just inaccurate. It's a misalignment between vendor and customer that neither side can afford.
The companies that build their revenue architecture around outcomes---not headcount---will capture more value, retain customers longer, and invest more aggressively in the intelligence that produces those outcomes. The proxy served its purpose. It's time to retire it.