What is Seat Utilisation?

Seat utilisation is the only productivity metric that improves when the office shrinks. That alone should settle what kind of metric it is. Yet in outsourced delivery it still steers decisions it has no business steering.

A facilities ratio in a productivity costume

Seat utilisation measures how fully physical workstations are used: seats occupied as a share of seats available, over a day, a week or a lease term. Multi-shift operations extend the idea, so a desk serving two shifts counts twice, and the metric is sometimes expressed as shifts per seat. As a facilities question this is legitimate and useful. How much building does the operation need? When does the lease start constraining growth? Is the second shift paying for the property? Facilities cost is real, and someone should manage it.

The trouble begins when the number migrates from the property review into the delivery review. A seat is not output. A full floor is not a performing operation. The metric certifies that the building is busy, and nothing else, but because it is visible, simple and photogenic, it gets read as evidence of productivity by people who would never accept that logic anywhere else.

Seats, people, output

Three layers sit under the number, and they move independently.

• A seat can be full while its occupant produces nothing that earns. Occupied is not productive: an agent idle between contacts fills a seat exactly as well as one handling them, and a team kept on-site to make the floor look committed fills it best of all.
• People can produce without seats. Remote and hybrid work detached output from the building years ago, and automated handling detaches it further. An operation can grow its output while emptying its floors.
• Output can rise as seats fall. Better routing, deflection of simple contacts and higher first-contact resolution all improve the service while making the building look worse.

Each layer has its own instrument. Utilisation rate measures how a person’s paid time is spent. Occupancy measures how relentlessly an agent’s logged-in time is consumed by work, and it misleads in its own particular ways. Seat utilisation measures the furniture. Confusing the three means managing whichever is easiest to see, and the building is always the easiest to see.

How per-seat economics distort delivery

In outsourced delivery the seat is often the unit of commerce: contracts priced per seat per month, sites judged on seat fill, growth reported in seats sold. Once revenue attaches to filled seats, the metric acquires gravity, and it bends decisions toward the building. Automation becomes revenue risk, because anything that deflects contacts empties seats: the commercial engine votes against improvements the client would value. Schedules drift toward what fills shifts around the clock rather than what matches the demand curve. And headcount quietly becomes the product: when the seat is what is sold, filling it counts as success regardless of what the person in it produces, while training and coaching time read as cost against the seat.

For illustration only: a site of five hundred seats run at high fill across two shifts reports excellent seat utilisation even if half the people in those seats sit idle between contacts. The building is busy. The work is not. The number cannot tell the difference, because it was never designed to.

One concrete example

Clearly illustrative, with no customer implied. An outsourcer prices a long-running support contract per seat. The client’s product improves, and chat plus self-service deflect the simple contacts. The right delivery decision is obvious: fewer, better-trained people handling the harder contacts that remain. But every step of that decision reduces billable seats, so the account team defends the floor instead. Deflection is slow-walked, headcount is held, and investment goes to seat fill rather than skill. Quality on the hard contacts sags, and at renewal the client moves to a provider selling outcomes rather than furniture. The seats emptied anyway. The pricing model just chose the slowest and most expensive route there.

The decision-intelligence angle

Seat utilisation answers a facilities question, and every time it is allowed to answer a delivery question, the building is making the decision. The remedy is not to abolish the metric but to return it to its lane, and to make the collisions explicit. When a delivery improvement collides with per-seat economics, that collision is a decision, and in a decision-intelligence view it is treated as one: it has an owner, the options are priced in cost-to-serve and outcome terms rather than seats, and the choice is recorded rather than resolved by default. Metrics are instruments, and decisions deserve to know which instrument they are reading. A firm that can say who decided to keep the floor full has at least made a decision. A firm that cannot has let its lease make it.