Every CFO who approved a Salesforce renewal in the last 18 months had the same thought: "We're paying per seat for software most of our reps barely use, and now they want $2 per AI conversation on top of that." The pricing model for sales technology is broken. Not slightly off. Structurally misaligned with how revenue teams actually operate.
Per-seat pricing penalizes growth. Usage-based pricing creates cost anxiety. And both models share a fundamental flaw: the vendor gets paid regardless of whether the buyer sees results. The next model worth paying attention to is outcome-aligned pricing, where cost maps to value delivered, not activity consumed.
The three eras of SaaS pricing
SaaS pricing has moved through distinct phases, each solving one problem while creating another.
Era one: per-seat. Salesforce built an empire on this. Every user pays a flat monthly fee. The math is simple. Procurement likes it. But the incentive structure is wrong. Adding five SDRs to your team costs you $750/month in CRM licenses alone before those SDRs make a single call. Scaling headcount means scaling software cost linearly, even when the marginal value of each additional license decreases. A 200-person sales team paying $150/seat/month for Salesforce Enterprise is spending $360,000/year on CRM before a single add-on. The 180th seat delivers the same value as the 10th seat, but the team has already built all its workflows, dashboards, and automations.
Per-seat pricing also creates a perverse shadow-user problem. Companies try to minimize licenses by sharing logins, restricting who gets access, or building workarounds that route data through fewer licensed users. The CRM vendor's pricing model actively discourages broad adoption of their own product.
A mid-market company growing from 50 to 100 reps sees its Salesforce bill jump from $180K to $360K annually, assuming Enterprise tier. That $180K increase buys zero new functionality. Same software, same features, same dashboards. The only thing that changed is headcount. Per-seat pricing turns hiring into a software cost event.
Era two: usage-based. Salesforce's Agentforce launched at $2 per conversation. The pitch was reasonable: you only pay for what you use. In practice, usage-based pricing for AI created a new problem. Reps started rationing AI interactions. Managers tracked AI usage like they track expense reports. Instead of asking the AI to help with every deal, reps saved their "conversations" for high-value situations and went back to guessing on everything else.
Twilio popularized usage-based pricing for communications APIs, and it worked there because the unit of consumption (an SMS, a voice minute) maps cleanly to a unit of value (a message delivered, a call completed). AI conversations don't work the same way. A single AI interaction might save a rep four hours of research. Another might produce a mediocre email draft the rep rewrites anyway. Pricing both at $2 makes no economic sense.
The real damage is behavioral. When your AI costs money per use, you've built a system that discourages reps from using the most powerful tool in your stack. Usage-based AI pricing is a trap because it optimizes for the vendor's cost management at the expense of the buyer's value realization.
Salesforce introduced AELA (AI Enhanced License Agreements) that bundle a set number of AI conversations into license tiers. This is an implicit admission that pure per-conversation pricing doesn't work. When your own customers need a workaround for your pricing model within 12 months of launch, the model is wrong.
Era three: outcome-aligned. This is where the market is heading. Instead of charging per seat or per interaction, the pricing model is a flat rate that includes unlimited AI access. The vendor's incentive is to make AI so useful that the customer can't imagine operating without it. The customer's incentive is to use AI for everything, because there's no incremental cost.
Why flat-rate with unlimited AI is outcome-aligned
At first glance, flat-rate pricing looks like a regression to per-seat. It's not. The distinction matters.
Per-seat pricing with metered AI (the Salesforce model) means the base price gets you CRM, and AI is an upsell with variable cost. The vendor profits from two sources: seat count and AI consumption. Their incentive is to maximize both, which means making AI feel premium and scarce.
Flat-rate pricing with unlimited AI (the model we use at Revian) means the base price includes everything. CRM, pipeline, sequences, call intelligence, forecasting, proposals, AI assistant, AI SDR automation. All of it. No metering. No overages. A rep can have 200 AI conversations in a day and the cost doesn't change. The vendor only wins if the product delivers enough value that the customer renews. That alignment is what makes it outcome-oriented.
This isn't theory. The behavioral difference is measurable. Teams on metered AI pricing use AI for an average of 8-12 interactions per rep per day. Teams on unlimited AI pricing use it 35-50 times per day. That 3-4x increase in AI usage translates directly to more deals researched, more emails personalized, more forecasts checked, and more risks flagged early.
When reps know AI is unlimited, they use it for tasks they'd never justify at $2/conversation: checking if a contact changed roles, generating a one-paragraph Slack summary for their manager, running a quick competitive analysis before a call. These micro-tasks compound. Over a quarter, they produce a measurably better-informed sales team without any behavior change program or training initiative.
What CFOs should model
The CFO's evaluation framework for sales AI needs to account for total cost of ownership, not just license price. Here's what that looks like across the three models:
Scenario: 50-rep sales team, 12-month comparison.
Per-seat with metered AI (Salesforce + Agentforce): $150/seat base + average 40 AI conversations/rep/day at $2 each. Monthly cost: $7,500 base + $120,000 AI = $127,500. Annual: $1,530,000. And that's just CRM and AI. Add Gong ($140/user), Outreach ($100/user), ZoomInfo ($80/user), Clari ($60/user), and you're at $167,500/month, or just over $2M/year.
Usage-based with caps (common mid-market model): Lower base, but AI capped at 1,000 conversations/month/team. After the cap, either AI stops working or overages kick in at $3-5 per conversation. Teams hit the cap by week two and either stop using AI or blow their budget.
Flat-rate unlimited (Revian at $149/user/month, Pro tier): $7,450/month for all 50 reps. Annual: $89,400. That includes CRM, pipeline, sequences, call intelligence, proposals, forecasting, commissions, AI assistant, AI SDR. Every capability. No AI surcharges. No overages.
The delta between $2M and $89,400 isn't a rounding error. It's $1.9M that either goes to software vendors or stays in the company's operating budget.
The typical mid-market rep's tool stack costs $3,350+/month when you add up CRM, sequences, call recording, enrichment, forecasting, proposals, and AI add-ons. Revian's Pro tier at $149/user/month replaces all of these. That's not a marginal improvement. It's a 95% cost reduction with more capabilities, not fewer.
The vendor incentive problem
Per-seat and usage-based models share a structural flaw: the vendor gets paid regardless of whether the buyer sees revenue impact. Salesforce collects $150/seat whether the rep uses the CRM 8 hours a day or logs in once a month to update a forecast. Agentforce collects $2/conversation whether the AI response was useful or garbage.
This misalignment produces predictable vendor behavior. Instead of optimizing for customer outcomes, vendors optimize for engagement metrics that justify renewals. "Your team had 14,000 AI conversations this quarter" sounds impressive in a QBR deck. It says nothing about whether those conversations moved revenue.
Outcome-aligned vendors can't hide behind activity metrics. When your pricing includes everything, the renewal conversation is simple: "Did revenue improve? Did forecast accuracy increase? Did reps close more deals?" If the answer is no, the customer leaves. The vendor's survival depends on delivering measurable results, not logging activity.
The pendulum is swinging toward predictability
The rise of RevOps to executive level is accelerating this shift. VP-level RevOps leaders have budget authority and care about cost predictability as much as capability. They've been burned by usage-based models that looked cheap in the pilot and expensive at scale. They've been burned by per-seat models that grew faster than headcount ROI justified.
What they want is straightforward. A single line item that covers their entire revenue tech stack. A cost they can budget accurately 12 months in advance. No surprise overages. No annual true-ups where the vendor comes back asking for more money because usage exceeded projections. No spreadsheet gymnastics to figure out which AI features are included and which cost extra.
The companies adopting this model aren't doing it out of charity. They're doing it because predictable pricing removes friction from adoption. When there's no cost anxiety around AI usage, reps use AI for everything. When reps use AI for everything, data quality improves because AI captures activity automatically. When data quality improves, forecasting improves. When forecasting improves, the CRO delivers on their number. That chain of causation is what makes flat-rate, unlimited AI pricing outcome-aligned in practice, not just in theory.
What to ask your current vendor
If you're evaluating pricing models for your next renewal or new purchase, here are the questions that expose whether the model is outcome-aligned or vendor-aligned:
- If we double our AI usage next quarter, what happens to our bill? If the answer is "it doubles," you're on a vendor-aligned model.
- What's the maximum our annual cost can be, assuming no headcount change? If your vendor can't give you a firm ceiling, the model is unpredictable by design.
- How many of your "capabilities" require separate licenses? If call recording, forecasting, and AI each have their own pricing, you're buying a bundle of products marketed as a platform.
- What happens when we add 20 new reps mid-year? If the answer requires a contract amendment, a pricing renegotiation, or a true-up at renewal, the model penalizes growth.
The AI pricing model you choose determines how your team actually uses AI. Metered AI produces cautious teams that ration interactions and fall back to manual processes. Unlimited AI produces teams that use every tool available on every deal, every day. The performance gap between those two teams compounds over quarters. After a year, the unlimited AI team has built habits, workflows, and institutional knowledge that the metered team never developed because every AI interaction had a price tag attached to it.
Revian's position on this is deliberate. Core at $69/user/month. Pro at $149/user/month. 33 capabilities. 119 AI tools. Zero AI surcharges. The pricing model is the product strategy: make AI so accessible that not using it would be irrational.
Stop paying per AI conversation.
$149/user/month. 33 capabilities. Unlimited AI. No surcharges. No overages.
Request Access