Two attendance vendors quote you. One says €2 per employee per month. The other says €400 per device, once. Which is cheaper? You cannot tell — not without knowing how many doors you have, how fast you hire, and whether you will still be running the system in five years. Pricing model matters more than price. Here is how each one behaves as your organisation changes.
Per employee, per month
The dominant cloud model. You pay a small amount for every person enrolled, billed monthly or annually, and check-in devices are unlimited or generous.
Behaves well when: you have many employees relative to entry points — offices, hospitals, schools, retail chains. Adding a second or third check-in point costs nothing, so you can put a tablet in the canteen and another at the loading bay without a budget conversation.
Behaves badly when: you employ large numbers of seasonal or rotating staff. If billing counts every person enrolled rather than active employees, harvest season can triple your bill. Ask precisely how the vendor counts: enrolled, active this month, or peak in the period.
What to negotiate: the definition of an active employee, annual-billing discount (commonly around 20%), and whether the count resets when seasonal staff leave.
Per device or per terminal
Typical of hardware-led vendors. You pay for each physical check-in point, sometimes as a one-off, increasingly as an annual fee per device.
Behaves well when: you have a large workforce funnelling through very few gates. A thousand people through two turnstiles is the ideal case — the per-head cost approaches zero.
Behaves badly when: you want convenience. Every additional check-in point has a price, so teams ration them: one terminal for the whole site, queues at shift change, and employees walking five minutes to clock in. It also discourages using phones as check-in devices, which is the cheapest capacity you have.
Watch for: device fees introduced at renewal. Several cloud attendance products have moved previously free device slots behind annual per-device charges, which converts a fixed cost into a growing one. Ask whether device pricing is contractually fixed for the term.
Perpetual licence plus maintenance
You buy the software outright, usually sized by employee count, and pay an annual maintenance fee — commonly around a fifth of the licence value — for updates and support.
Behaves well when: you have capital budget rather than operating budget, you intend to run the system for many years, and you are self-hosting anyway. Public-sector and defence buyers often prefer it for exactly these reasons. Over a seven-year horizon it frequently beats subscription.
Behaves badly when: headcount grows unpredictably, or you want to change your mind in year two. You have bought a fixed size; expanding means another purchase.
What to check: whether maintenance is optional (and what you lose if you decline), and whether the licence covers the on-premise server or only clients. Our on-premise attendance software page covers how self-hosted licensing usually works.
Freemium and free tiers
A free plan capped by employee count, designed so small teams can run permanently and larger ones can trial properly. It is the lowest-risk way to validate that recognition works on your devices, in your lighting, with your people, before any procurement conversation. NCheck’s free plan covers five employees indefinitely and includes face check-in, GPS and reports.
The thing to verify is what the cap actually restricts. A tier that limits employees is a genuine trial. One that disables liveness detection, offline mode or reports is a demo with a countdown.
Metered extras
Some capabilities are billed by use rather than by seat. Visitor check-ins are the usual example — charged per appointment rather than per receptionist — and per-device app licences for administrator clients are another. These are not inherently bad; metered pricing is often fairer for occasional use. They simply need to be in your model. See the pricing page for how NCheck separates these lines.
Comparing models fairly: the three-number method
Ask every vendor for exactly three figures, and refuse to compare on anything else:
- Year-one total, including hardware, implementation and any one-off fees.
- Year-two total, assuming no growth. This strips out introductory discounts and reveals the real run rate.
- The delta for adding one site, one check-in point and fifty employees. This exposes which model punishes the growth you are actually planning.
Put the three numbers side by side and the right model usually becomes obvious within minutes — far faster than arguing about per-unit rates.
Which model suits which organisation
- Under 50 employees, one location: per employee, or a free tier. Avoid anything with per-device fees.
- Manufacturing or logistics, many gates: per employee, provided devices are unlimited. Per-device pricing is where these sites overspend.
- Seasonal workforces: per employee, but only with a clear active-user definition. See agriculture for how peak-season counting plays out.
- Government, defence, regulated sites: perpetual plus maintenance, self-hosted. Procurement rules often require it, and our compliance hub explains the data-residency reasoning.
- Multi-country groups: per employee with volume tiers, plus a serious look at where data must reside in each jurisdiction.
Frequently asked questions
Is annual billing always worth it?
If you are confident in the product, yes — discounts around 20% are standard. Run a monthly pilot first, then switch to annual once accuracy and adoption are proven.
Can we move from cloud subscription to a perpetual on-premise licence later?
Only if the vendor runs the same server software in both modes. Where it does, you can pilot in the cloud and move on-premises without re-enrolling employees. Where it does not, you start again.
Are per-device fees ever the better deal?
Yes — very large headcounts passing through a handful of fixed gates. Outside that pattern they tend to grow faster than headcount-based pricing.
Contract terms that matter more than the rate
Once you have chosen a model, the protection comes from the contract rather than the price list. Four clauses are worth more than a discount.
Price-lock and uplift cap. Fix the per-unit rate for the initial term and cap the increase at renewal. An uncapped renewal on a system holding your biometric enrolments is a weak position, because switching means re-enrolling everyone — your practical leverage at renewal is lower than it feels at signature.
No new billable units mid-term. State that units not billable today — devices, sites, administrator seats, API calls — cannot become billable during the term. This is the single clause that would have protected buyers caught by per-device fees appearing at renewal.
Data export on exit. Require that attendance history is exportable in a documented, non-proprietary format at any time, and that templates are deleted on termination with written confirmation. Historical hours are often needed for years after you stop using a system.
Down-scaling, not just up. Most contracts let you add employees easily and reduce them never. If your headcount is seasonal or uncertain, negotiate the ability to reduce the committed count at renewal.
Watch the total, not the unit
A useful sanity check before signing: divide the annual total by the number of employees who will actually clock in, and compare that single figure across vendors. It collapses every pricing model into one comparable number and consistently exposes quotes that looked competitive per unit but arrive somewhere else in total. Run it again with next year’s expected headcount and site count — the vendor whose number barely moves is the one whose model fits your growth.
NCheck is a biometric attendance system by Neurotechnology that runs on-premises or in the cloud, supports face, fingerprint and iris recognition, and works on phones, tablets, IP cameras and biometric terminals. Free forever for up to 5 employees.
