They win the argument. They lose the sale.

Software stocks shed close to two trillion dollars this year. The market had done the maths on seats. Every AI-native founder saw an opening. It is one. Just not the one they think.

M1 The fortress

Thirty years of software billed per human. Easy to forecast, approve and compare. Do not attack it. Map it, then leave it.

M2 The dependency

Revenue, forecast, share price, sales compensation. All of it sits on a count of licensed humans.

M3 The fragility

Every agent that works puts a seat in question. The incumbent cannot grow its AI without weakening its own billing unit.

M4 The motivation

The incumbent fights harder to keep the seat than to replace it, because the seat is what its revenue is built on. You have no seat to defend. That decides who blinks.

M5 The economics

Discount one enterprise seat and every renewal in the base asks for the same. You price one outcome for one buyer. Every defence costs them more than it costs you.

M6 The fork

Reprice on outcomes and put your revenue base at risk, or hold the seat and watch it erode. Bain checked over thirty incumbents last autumn. Two thirds bolted an AI meter onto the seat. The rest raised seat prices. Nobody chose. They hedged.

Six moves in, you have the opening. This is where most challengers stop.

M7 The customer

Will the buyer move? Not yet. Hours saved are not budget released. Bain sees customers paying for both the people and the agents while budget crawls from labour lines to software lines. And procurement likes the seat for the same reason the incumbent does: it fits in a budget line.

Beating the incumbent is strategy. Getting the buyer to move is commercial architecture: who signs, what they must justify, what number fits where.

A pipeline full of enthusiastic pilots that never convert is not a sign you lost to the incumbent. It is a sign the last move was skipped. Fix the buyer's maths before you fix your pricing page.