The headline overstates it. SaaS is not dying and monthly billing is not going anywhere. What is genuinely breaking is the pricing metric, and specifically the seat.
Why the seat worked
Per-seat pricing survived twenty years because it satisfied three conditions at once:
- It tracked value. More people using the tool meant more work getting
done through it.
- It was legible. A buyer could forecast next year's bill from next year's
headcount plan.
- It was cheap to meter. Counting logins is trivial.
That is a strong combination, and it is why seats beat every proposed alternative for two decades.
What AI breaks
Condition 1, and it breaks in the worst possible direction.
When the software performs the work rather than assisting a person performing it, seats stop correlating with value delivered and start correlating inversely with it. A tool that lets a team of six do what eight did is a tool whose own bill should fall. The vendor is now penalised for the exact outcome it sells.
That is not a tension a discount fixes. It is a structural conflict between the price metric and the product's purpose.
The three replacements, and what each costs
Usage pricing. Bill by the unit of work performed: messages sent, records enriched, calls placed, tokens consumed.
- Works because it tracks value again, and it scales down for small buyers.
- Costs forecastability. Finance hates a variable bill, and the fear of a
surprise invoice suppresses exactly the usage the vendor wants. The usual patch is a committed floor with overage, which is really a hybrid.
Outcome pricing. Bill per meeting booked, per qualified opportunity, per resolved ticket.
- Works because it is the most honest alignment available, and it is
extremely easy to sell.
- Costs attribution, which is the real barrier. Charging per meeting means
agreeing on what counts as a meeting, who caused it, and what happens when a human and the system both touched it. Most companies cannot answer that about their own pipeline today, which is a measurement problem long before it is a pricing one. It also puts vendor margin at the mercy of the buyer's execution.
Hybrid: platform fee plus consumption. A base for access and support, a variable component for work performed.
- Works because it keeps a forecastable floor and restores value tracking
above it, and it is what most of the market is actually converging on.
- Costs simplicity, and it needs a genuinely usable usage dashboard or every
renewal becomes an audit.
What does not change
- Recurring revenue does not die. Consumption revenue that recurs is still
recurring revenue; it just has different retention mechanics. Expansion moves from seat growth to usage growth, which shows up in net revenue retention rather than in headcount.
- Land and expand does not die. It changes unit.
- The buyer still wants a predictable number. Every successful consumption
model in market ships with commitments, caps, or alerts for exactly this reason.
What this means if you sell into GTM teams
The pricing conversation is now a measurement conversation. You cannot charge per outcome unless both sides can see the outcome and agree on its definition, and that is a data problem your buyer probably has not solved.
The practical read: the vendors moving to outcome pricing successfully are the ones who shipped the measurement layer first, then priced against it. The ones announcing outcome pricing without it end up in quarterly disputes about attribution.
Where this sits
This is Measurement in the Tenbound Pipeline Architecture Standard. The pricing model a market can support is downstream of what that market can measure, which is why the shift is slower than the announcements suggest.
Related: ARR, net revenue retention, and AI in sales development.