The Old B2B Pricing Model Is Dead. These Are the New Kings
1. Key Themes
Theme 1: Per-seat pricing is eroding because AI agents decouple work from headcount
Seats stop measuring work when agents do the work
A seat only counts who can log in, so as AI makes teams more efficient, per-seat vendors earn less from their own product's success.
"When a finance team closes its books in 2 days instead of 10, or a support team resolves most tickets without a human, the work requires fewer logins. So a vendor charging per seat earns less precisely because its own product becomes more efficient."
Seat-based pricing is fading, and nobody knows where it lands
The direction is clear, but the destination is not, even to senior operators.
"Henry Schuck, the CEO of ZoomInfo, recently said that after speaking to hundreds of consultants and customers, nobody really knows where B2B pricing will end up, and the answer changes on a weekly basis at this point."
Buyers are squeezed, making seat renewals the first target
Spend is rising without more tools, and AI is being funded by cannibalizing existing software budgets.
"Zylo's 2026 SaaS Management Index shows average SaaS spend per organisation up 8% to $55.7 million, while the number of applications barely moved at 305, so companies are paying more for the same tools."
"A Redpoint survey of 141 CIOs, as reported by SaaStr, found that 45% are funding AI directly out of existing software line items, and 54% are actively consolidating vendors."
Theme 2: Three "pay-for-work" models are emerging, each with a distinct risk profile
Usage-based pricing: scales with value, but is hard to forecast
"Usage-based pricing charges customers for what they consume, the way electricity does. When value grows with volume, the price grows with it."
"The biggest issue with usage-based pricing is that it is hard to forecast, unless there is a pre-agreed spending cap, alerts, or other mechanisms that let customers control usage and spend."
Resolution-based pricing: the easiest pay-for-work model to adopt
"Both sides of the equation see the same number, so there is no pricing argument to be made here, and the customer pays only when the job is done. Also, nothing downstream has to be attributed, which makes this the easiest of the pay-for-work models to adopt."
The catch is vendor-side cost exposure:
"for vendors, it's advisable to know their own cost per resolution, because while the price may be fixed, the inference cost is not."
Outcome-based pricing: the highest alignment and the highest risk for vendors
"From the buyer's perspective, a software license turns into a business case and you only pay when it works. Some might call this model an 'underwriting business' because the whole risk rests on the vendor."
"proving that a certain product caused the result can be messy and ambiguous. And a spectacular result could mean a spectacular bill."
Theme 3: Hybrid pricing is becoming the default
Hybrid has overtaken every other model in one year
"Hybrid is already the most popular B2B pricing model at 37%, up from 25% twelve months earlier, according to Kyle Poyar's Growth Unhinged 2026 survey on B2B monetization."
Investors reportedly prefer hybrid over seats by a wide margin
"Respondents also say investors prefer hybrid (35%) far more than seat-based pricing (5%)."
Hybrid structure: a base fee plus a metered layer
"A hybrid usually implies a base fee with a metered layer on top."
Theme 4: Incumbents are structurally locked into seats; startups have a clean-sheet advantage
Seat pricing is wired into every incumbent function
"Billing counts seats, salespeople are paid on contract value booked upfront, forecasts assume smooth recurring revenue, and customer success is judged on renewals."
"It's practically impossible to move to resolution-based or outcome-based pricing and change everything at once without turning the business upside down."
The result: AI add-ons bolted onto seat contracts
"that's exactly why so many announced pricing changes end up as an AI add-on sold on top of the same seat contract."
Startups can price from scratch
"A startup doesn't have anybody to convert, so the first contract can be priced however the founder wants."
"Remember, pricing is part of the product. So work out what you are really selling, and build the metering and trust around it. That's the advantage an incumbent may never have."
2. Contrarian Perspectives
Incumbents aren't slow or broken; they're structurally wired for seats
The conventional take is that big SaaS vendors are sluggish or have bad models. The article argues the problem is organizational plumbing, not competence.
"Most people would say that big SaaS companies are slow or their models are faulty, but in reality, all of them are wired around seat pricing."
Supporting evidence: sales comp, billing, forecasting, and customer success metrics are all built on seat logic, and a seat-priced company "is responsible for giving its customer access. It's not responsible for whether the product worked as intended."
Efficiency gains can hurt the vendor under seat pricing
The perverse insight is that the better the AI product performs, the less a seat vendor earns.
"a vendor charging per seat earns less precisely because its own product becomes more efficient."
Seats are far from dead among large companies, and most startups should hedge
Despite the "dead" headline, the data shows seat pricing persisting at scale, and the author recommends flexibility over dogma.
"Although Poyar's survey has shown that 29% of B2B companies above $150M ARR still rely on seat-pricing, the majority of overall respondents are actively rethinking pricing for AI."
"You also don't have to pick one model forever, as 29% of companies in Poyar's survey already let customers choose between several."
3. Companies Identified
- Description: Enterprise software and cloud giant.
- Why mentioned: Its CEO predicted AI would force a rethink of software pricing, which the author says proved correct.
- Quote: "Satya was considering a 'per agent' and 'per user' pricing model at the time, since more and more work is being automated by AI agents."
- Description: B2B data and go-to-market software company.
- Why mentioned: Its CEO's view illustrates how unsettled B2B pricing is.
- Quote: "nobody really knows where B2B pricing will end up, and the answer changes on a weekly basis at this point."
Stripe, Twilio, Snowflake, Databricks, MongoDB
- Description: Infrastructure and data platform companies.
- Why mentioned: Cited as long-standing examples of usage-based pricing.
- Quote: "This is the model that companies like Stripe, Twilio, Snowflake, Databricks, MongoDB and others have been using for a while."
- Description: Enterprise CRM and SaaS incumbent.
- Why mentioned: The case study for an incumbent too large to easily leave seat pricing, now offering multiple payment options for its AI product.
- Quote: "The model is pretty hard to change once you reach a size of Salesforce's magnitude." (Chart caption: "Salesforce now sells Agentforce four ways.")
- Description: SaaS management platform and publisher of the 2026 SaaS Management Index.
- Why mentioned: Its data shows rising SaaS spend without more apps.
- Quote: "average SaaS spend per organisation up 8% to $55.7 million, while the number of applications barely moved at 305."
- Description: Venture capital firm.
- Why mentioned: Its CIO survey shows AI is being funded from existing software budgets.
- Quote: "45% are funding AI directly out of existing software line items, and 54% are actively consolidating vendors."
- Description: Outsourced finance and back-office services for VC-backed startups (sponsor).
- Why mentioned: Sponsor placement tied to the point that pricing changes affect runway and board scrutiny.
- Quote: "Graphite keeps the back office of VC-backed startups ahead of their next raise." / "1,000+ venture-backed companies supported over ten years"
4. People Identified
- Description: CEO of Microsoft.
- Why mentioned: Predicted that AI would reshape software pricing.
- Quote: "Satya Nadella appeared on the Dwarkesh Podcast saying that AI will fundamentally alter software pricing."
Henry Schuck
- Description: CEO of ZoomInfo.
- Why mentioned: Offers a candid view on how unsettled B2B pricing currently is, based on conversations with hundreds of consultants and customers.
- Quote: "after speaking to hundreds of consultants and customers, nobody really knows where B2B pricing will end up."
Kyle Poyar
- Description: Author of Growth Unhinged and its 2026 B2B monetization survey.
- Why mentioned: His survey provides the key adoption data on hybrid pricing, investor preference, and incumbent behavior.
- Quote: "Hybrid is already the most popular B2B pricing model at 37%, up from 25% twelve months earlier, according to Kyle Poyar's Growth Unhinged 2026 survey on B2B monetization."
Ruben Dominguez
- Description: Author of The VC Corner.
- Why mentioned: Writer of the piece; his prior work on SaaS durability and financial modeling is referenced.
- Quote: "Don't Believe Those Who Say that SaaS is Dying"
5. Operating Insights
Name your one countable unit before anything else
Pricing starts with identifying the single thing your customer would pay for on its own.
"start by naming the one main countable thing your product does. A thing that your customer would pay for on its own. This is a huge test. If you can name one, that is your unit."
If you can't, default to usage:
"If you can't name this one thing, then you should think about usage-based pricing."
Define "resolved" in writing and track unit cost from day one
Fixed per-resolution prices against variable inference costs can create unprofitable deals.
"You should also write the definition down before you sign your first customer, and track what one unit costs you to deliver from day one. Without that, a per-resolution price can win you deals that may lose money."
Pilot small and assign revenue recognition ownership early
Expect smaller initial contracts and accounting complexity from variable revenue.
"Test with a small group first and expect smaller first contracts, and decide early who owns revenue recognition, since variable revenue gets uncomfortable in an audit."
6. Overlooked Insights
AI inference costs are pushing the giants themselves toward metering
The author notes in passing that AI adds a cost layer for software vendors, which may force even large players to meter usage because they can't absorb unbounded costs. This is a margin-structure point that shapes pricing independent of buyer preference.
"AI is also adding an additional cost layer to the SaaS vendors, and the giants are already saying that when the cost goes beyond what they themselves can afford, they need to have some kind of meter themselves."
Letting customers choose their pricing model is already common
Offering multiple pricing options (as Salesforce does with Agentforce) is a quiet trend that lets vendors learn which model buyers prefer without committing to one.
"29% of companies in Poyar's survey already let customers choose between several."