Per-seat pricing in a world where seats are evaporating.

Pricing ChallengeWhy the per-seat model breaks. Three structural pressures.

Per-seat pricing was the cleanest unit-economic story in software history. One human, one license, one number on the contract. The motion that grew around it — sales-led growth, expansion through headcount and churn measured in seats was the most legible motion B2B has ever run.

It is also the motion least suited to the migration of work into agents.

The model breaks under three pressures simultaneously, which is why most attempts to patch it fail. Each pressure on its own is recoverable. The three of them in compound require a redesign, not a patch.

01. The buyer's workload is no longer human-bound.

Agents absorb hours that used to be billed against seats. A single licensed user, augmented by two or three agents, now produces work that previously required a team. The seat count stays flat or shrinks. The output curve bends upward. The buyer feels this first, as productivity. The seller feels it second, as expansion stalling. By the time the seller diagnoses what changed, the renewal is already in procurement.

02. The buyer's procurement agent re-asks the math.

What looked like reasonable per-seat economics now looks like over-collection against thinning logins. Procurement is no longer running a one-time price check at renewal. With AI in the buyer's stack, the math is re-asked every quarter — sometimes by a human, increasingly by an agent that pulls login data, compares it to seat count, and flags the gap automatically. The seller is no longer negotiating against a procurement cycle. The seller is negotiating against a continuous audit.

03. The seller's cost base is no longer headcount-linear.

Compute, inference, retrieval. The cost curve has decoupled from the price curve. Under per-seat economics, marginal cost of a new seat was near zero. Under agent-driven workloads, the cost of serving a single power user can exceed the cost of serving ten passive ones. Revenue scales with seats. Cost scales with work. Margin compresses without anyone deciding it should.

These three pressures don't show up sequentially. They show up at the same time, in the same renewal cycle, in the same procurement re-read. That is why the patch attempts fail — adding a usage-based add-on, raising seat price, throttling agent calls. Each one addresses a single pressure while the other two keep working in the dark.

The shift. The unit mismatch and its blast radius.

The diagnosis is two-part. The first sentence names the structural defect. The second predicts the damage.

Historically, one seat roughly equaled one person's output. Charging per seat was a clean proxy for charging for value. That equation has snapped. A single licensed user can now drive ten or a hundred times the output by running agents alongside them. The work being produced no longer lives inside the seat. The thing being billed — a login, and the thing being delivered, work product have come unglued.

Once those two units decouple, the damage isn't isolated to the price line. It propagates through every downstream system that was calibrated on the assumption that seats and work moved together.

• Revenue forecasting leaks. Expansion no longer follows headcount. The model that predicted Q+2 ARR off seat adds no longer maps to anything real.
• Sales comp leaks. Reps are compensated on seats added, not value delivered. The motion the comp plan reinforces is now the motion that misreads the market.
• Renewals leak. Procurement now does the math — or its agent does — and sees over-collection against thinning logins. The conversation that used to be a quiet auto-renewal becomes an opening to renegotiate the floor.
• Cost structure leaks in the opposite direction. Inference and compute scale with workload. Revenue still scales with seats. Margin compresses silently because the leak is internal — it never shows up on the customer's bill.
• Customer success leaks. Health is still measured on logins that no longer correlate with usage. A green account on the seat-login dashboard can be a red account on the work-volume dashboard, and CS doesn't see the second dashboard yet.

Leak is the operative word. It is not a single broken thing the team can patch. It is value escaping continuously, in small amounts, in every direction at once. The team cannot seal one leak without re-plumbing the motion that grew around per-seat economics.

Three conversion pathsConsumption, outcome, hybrid.

Three viable paths back to a model that prices the unit of value. None of them is risk-free. All of them require sequencing.

Path A — Consumption. Price the workload.

Bill the unit of work: inferences, runs, documents processed, decisions made. Cleanest math. Hardest sales motion to reposition.

The math is clean because the unit is observable. Every API call, every job, every transaction can be counted, billed, and reconciled. The motion is hard because the buyer can no longer forecast their own bill — and the buyer's reaction to that uncertainty is the conversion's most common point of failure.

Buyers accept consumption pricing only when the seller absorbs the forecasting risk. That means commitments, floors, or visibility tooling that gives the buyer a credible monthly estimate ahead of time. The studio team has experienced this before and learned this lesson the hard way. The forecast layer has to be built in from the start.

Path B — Outcome. Price the result.

Bill the customer for the qualified pipeline, the recovered revenue, the headcount the system absorbed. Highest alignment. Hardest to underwrite contractually.

The alignment is real. When the buyer pays only on outcomes, the seller's incentive is fused to the buyer's value realization. The legal problem is real too. Underwriting an outcome contract requires agreement on attribution, baseline, measurement window, and dispute path. Most legal teams will not sign the first version. Most finance teams will not book the revenue under standard recognition until the path is proven.

Outcome pricing works in motions where the outcome is countable in days, not quarters — and where the seller has the operational discipline to instrument attribution from day one.

Path C — Hybrid. Reserved consumption floor with outcome-linked variability above.

The most common shape for future success. The shape that survives procurement.

The floor gives finance the revenue predictability they need to book and forecast. The variability above gives the buyer the upside-aligned story they need to defend the spend to their own procurement function. The hybrid is structurally more complex to operate, with three pricing dimensions instead of one. But the failure rate at procurement is dramatically lower because each constituency on the buy side gets the story they need.

The conversion is not a price change — it is a motion change. Sales has to re-sell. Comp has to re-design. CS has to re-instrument. Finance has to re-forecast. Product has to expose the metering surface area.

Business leaders that treat the pricing change as a contract amendment will likely fail. They need to treat it as a full GTM redesign to succeed — slowly, and with explicit sequencing of which constituency is migrated first.

The ObstaclesThe top five failure modes we see.

The studio team has run into all of them at various times. Our models are designed to avoid these modes, but they still exist. Each one can sink the conversion if not addressed head on.

The five modes, named here as preview:

1. The forecasting void. Buyers reject consumption because they cannot predict their own bill.
2. The comp plan collapse. Reps stop selling because the new model does not pay them on the motion they know.
3. The CS blind spot. The health dashboard still tracks seats and misses the real usage signal.
4. The procurement re-read. An existing customer renegotiates the floor on the strength of the new pricing logic.
5. The margin trap. Consumption is priced below the true compute cost curve at scale.

Field notes from those engagements will fill the rest of this brief in the next release.

05 - CloseThe pricing model is the proxy. The motion is the system.

Sitting on a pricing-model conversion is sitting on a motion redesign in disguise — and the longer the conversion sits, the more leaks the motion compounds.

If the motion is breaking on expansion gone flat, renewals getting re-read, or inference cost eating margin — the audit is the right way in.