The Empty Chair Discount: How UK Small Firms Can Pay for Software by the Desk, Not the Head

There's a particular invoice line that makes UK small-business owners twitch. Twelve people on the payroll. Four of them using the software at any one time. Twelve licences on the bill, every month, forever.

That gap between headcount and actual use is where concurrent licensing lives. It is not a loophole and it is not a grey-market trick. It is a legitimate licensing model that some vendors still sell, and for the right kind of team it can cut a software bill by a third without losing a single member of staff.

Here is how to work out whether it applies to you, which vendors actually offer it in 2026, and the traps that catch people out.

Two ways to count a user

A named-user licence is tied to a person. Sam has a seat, Priya has a seat, and if neither of them logs in for a fortnight you have still paid for both. Most modern SaaS works this way, and it is simple to administer.

A concurrent licence — sometimes called a network licence or floating licence — is tied to a pool of simultaneous users. You buy eight seats, and any eight people can be logged in at once. Person nine waits, or knocks someone off. The software usually sits on a licence server inside your office or, increasingly, on a vendor-hosted licence manager.

The attraction for a UK firm with shift patterns, part-time staff, field engineers or a sales team that is mostly on the road is obvious. The risk is equally obvious: buy too few seats and you have built a queue outside the tools people need to do their jobs.

The break-even maths, without the spreadsheet headache

The starting point is not headcount. It is peak concurrent usage — the highest number of people logged in at the same moment on a normal working day. Not the average. The peak.

Get that number from the licence server logs if you already run one, or from a week of asking people to shout when they log in. Then add a buffer of one or two seats for the days when everyone surfaces at once.

Now compare the two annual costs:

• Named: headcount × price per named seat
• Concurrent: (peak users + buffer) × price per concurrent seat

The catch is that a concurrent seat usually costs more than a named one. Vendors price in the flexibility. So the question is whether the gap in seat count is big enough to swallow the premium.

Take a twelve-person engineering consultancy. A named seat runs £1,200 a year. Peak concurrent use is four, so they buy five floating seats as a buffer. The concurrent seat is priced at £1,680 — a 40% premium. Five seats cost £8,400. Twelve named seats cost £14,400. Saving: £6,000 a year, before VAT.

Now change the peak to nine. Ten floating seats at £1,680 is £16,800. The named licence wins. The model only pays when the ratio of peak users to headcount is comfortably below the inverse of the price premium.

A rough rule for UK buyers: if peak usage is under 60% of headcount and the concurrent premium is under 50%, it is worth pricing properly. If peak usage is over 75%, stop and walk away.

Who actually sells concurrent licences in 2026

The list is shorter than it was a decade ago, and it is concentrated in engineering, scientific and design software.

Still broadly concurrent-friendly: SolidWorks network licences, Siemens NX, Vectorworks, MathWorks MATLAB, Minitab, Stata, Origin, and several Bentley and Trimble products aimed at architecture and surveying firms. These are the classic floating-licence markets, and resellers in the UK know them well.

Mixed: Microsoft still sells per-device CALs and Remote Desktop Services licensing alongside per-user options, and some on-premises Dynamics deployments can be licensed per user or per device. That is not the same as concurrency, but it is a useful lever for shift-based teams.

Named-only, despite what a grey reseller may tell you: Adobe Creative Cloud and Autodesk's current subscription products are named-user models. If someone offers you a 'network' Adobe or AutoCAD licence at a suspiciously low price, you are almost certainly looking at a shared login, a hacked key or a licence that will fail an audit.

Three traps that turn a saving into a liability

1. Cloud terms usually forbid it

Most SaaS vendors define a user as a human being with a login. Sharing one account between three people is not concurrent licensing, it is a breach of terms, and it is exactly the kind of thing that surfaces during a vendor audit. Concurrency has to be written into the contract, not improvised with a shared password.

2. Minimum seat counts and server fees

Some vendors set a floor — five seats minimum, for example — or charge separately for the licence server. Others bundle maintenance into the concurrent price but charge it annually whether you use the seats or not. Ask for the full three-year cost, not the headline seat price.

3. Audit evidence is your job

With named licences, the vendor can see who is logged in. With concurrent licences, the proof lives on your licence server. Keep monthly usage exports, a written internal policy on who may use which pool, and at least twelve months of logs. If an audit arrives, you need to show that peak usage never exceeded the seats you paid for. No logs, no defence.

The questions to ask before you sign

Before you commit to a floating model, put these to your UK reseller in writing:

• Is this genuinely concurrent, or is it a named licence with a shared login?
• What is the minimum seat count, and is there a separate licence-server charge?
• Can we export usage reports ourselves, and how far back do they go?
• What happens if we move to the vendor's cloud version — does concurrency survive the migration?
• What is the process if we need to burst above our seat count for a busy month?

The bottom line

Concurrent licensing is not a magic wand. It suits firms with predictable peaks, part-time or field-based staff, and a willingness to keep decent records. For a UK small business with twelve staff and four desks in use at once, it can be the difference between a sensible software bill and a monthly donation to a vendor's shareholder returns.

Do the peak-usage count first. Then do the maths. The empty chairs will tell you whether it is worth it.