Per-Seat Licensing Model

Companies today usually no longer buy software once on a CD, but rent it on an ongoing basis. This requires a rule for how much money flows. Under the per-seat licensing model, this rule is particularly simple: the provider counts the people who are allowed to use the program. For each of these people, the company pays the same fixed amount, for example 30 euros a month. “Seat” refers here to an access slot for exactly one person. Whether this person opens the program for eight hours daily or only once a quarter makes no difference to the price.

Why providers like billing by headcount

For the provider, this model is easy to plan. At the start of the year, they know fairly precisely how much money will come in. If the customer grows and hires people, revenue grows automatically along with it. In the financial world, such reliable, recurring revenue is called “subscription revenue,” and investors value it considerably more highly than one-off sales.

The model also has advantages for the customer. Costs are transparent and easy to explain: 200 employees times 30 euros equals 6,000 euros a month. Nobody has to measure how often someone clicks a button. That is precisely why this model has prevailed for decades in office software, video conferencing, and sales tools.

The catch becomes apparent when many accesses go unused. Large companies often pay for hundreds of licenses that hardly anyone opens. Experts call this “shelfware” — software that gathers dust on the shelf. During cost-cutting rounds, this is usually the first item the procurement department trims.

What a seat means technically

In practice, a seat is a user account with its own login. The provider maintains a list of these accounts and reconciles it against the contract. If the company wants to add one more person, an administrator books an additional seat. Billing is then usually adjusted proportionally for the remainder of the contract term.

It’s important to distinguish this from the so-called concurrent license. There, what counts is not who is registered, but how many people are working at the same time. Ten concurrent licenses can thus suffice for forty employees working in shifts. Per-seat is stricter: forty employees need forty seats, even if they never overlap.

Contracts also set minimum terms and lower limits. Often you can add seats at any time, but can only remove them again at the end of the contract term. A common misconception is therefore that the model is completely flexible. It is flexible upward, but rarely downward.

The dispute over seats in the AI era

Anyone who sees a subscription for an office suite, a video conferencing program, or a music service for teams is almost always looking at a per-seat model. AI assistants, too, were initially sold this way, for instance with a fixed surcharge per employee per month. That’s why the metric “revenue per seat” regularly appears in the quarterly reports of major software companies.

Since AI systems have started completing tasks on their own, the model has come under pressure. If a program takes over the work of three clerks, the headcount falls. Billed by seats, the provider would then earn less, even though its product delivers more. Several providers are therefore testing billing based on completed tasks rather than on people.

For investors, this is one of the most important open questions in the software sector. They are watching whether customers are cutting back seats and whether new pricing models make up for the shortfall. The shift has not been decided once and for all. Many companies currently combine both approaches: a base fee per seat plus an additional fee for actual usage.