Through a deliberate action or a deliberately short range - because Bluetooth alone reaches far enough to open doors people are only walking past.

Bluetooth Low Energy can be detected across many metres, which is useful for a vehicle gate and useless for a door onto a corridor.

Readers therefore offer several intent mechanisms: a tap on the reader itself, a gesture with the phone, a button in the application, or a very short range mode that behaves like a card.

Near-field communication is inherently short range and needs no intent mechanism at all, which is why many deployments prefer it where the handset supports it.

Range is adjustable in commissioning and is the setting most often left at default. It should be set per door.

Test it with the door closed and someone walking past at normal pace. A reader that triggers from the corridor will be switched off within a week.

A well-designed mobile credential works offline - it is stored on the device and presented locally, not fetched at the door.

This matters because doors are often in basements, lift lobbies and plant areas where signal is poor and the network is exactly what is missing.

The credential is provisioned over the network once, then held on the handset. After that the exchange with the reader is local.

What does need connectivity is provisioning a new credential and, in some products, revoking one - so a revocation may not take effect on the handset until it next connects.

Ask specifically how revocation is enforced offline. Some systems handle it at the controller instead, which is the more robust arrangement.

Confirm behaviour with the phone in battery-saver mode as well, which restricts background radio activity on some platforms.

Yes, in almost every case. Visitors, contractors and anyone without a suitable handset still need a physical credential.

A mobile-only door forces a parallel process for everyone outside the staff population, and those people are a significant fraction of the traffic in most buildings.

Combined readers cost little more and remove the need to maintain two kinds of door.

They also protect against the practical failures: a flat battery, a lost phone, a handset that will not run the application, or an employee who declines to install it on a personal device.

That last point deserves attention before rollout, because in many jurisdictions requiring an application on a personal phone is a matter for policy or consultation rather than a unilateral decision.

Keep a small pool of managed cards for these cases and enrol them the same way as any other credential.

Current versions of the major mobile platforms - but the supported version floor rises over time and older handsets fall off it.

Support depends on the credential vendor's application, not on the reader hardware, so the answer changes over the life of the installation.

Near-field communication availability differs by platform and by handset generation, and some platforms restrict access to it, which is why Bluetooth is the common denominator.

Ask for the current minimum operating system version and the vendor's policy on how long older versions are supported.

On a site with a wide range of personal devices, expect a tail of handsets that cannot run the credential at all.

Managed corporate devices simplify this considerably, because the estate can be kept within a known range of versions.

Generally yes - the credential is cryptographic, the device is already protected by a passcode or biometric, and revocation is immediate.

The exchange with the reader is authenticated in the same way as a smart card, so it is not copyable by recording it.

The phone adds a second factor almost incidentally: unlocking the handset is usually required before the credential can be used.

Revocation is the strongest advantage. A lost phone's credential is disabled centrally in seconds, with no card to recover.

The new risks are different rather than absent: the handset's own security posture, the vendor's cloud service, and the enrolment process by which a credential is bound to a device.

Ask how the vendor prevents a credential being moved to a second device, since that is the attack that has no equivalent in the card world.

Usually a recurring per-user subscription rather than a one-off purchase - so the comparison with cards is a total cost over the system's life.

Most vendors licence mobile credentials annually per user, sometimes with a multi-year option at a lower rate.

Against that, there is no card stock, no printer, no consumables and very little administrative handling.

On populations with high turnover the mobile model frequently wins, because the dominant cost of physical credentials is replacement and administration rather than the card itself.

On a stable population with low turnover, cards can be cheaper over five years.

Model both including labour, and confirm what happens to issued credentials if the subscription lapses - the answers vary and are rarely prominent.

It should be. Mobile credential platforms evolve, and a reader that cannot take a firmware update becomes obsolete faster than the door it is mounted on.

Firmware updates deliver new credential formats, security fixes and changes required by handset platform updates.

The best arrangement is central update over the supervised link from the controller, so an estate can be updated without visiting every door.

Configuration cards are the common alternative: a card presented to the reader carries the new settings. Workable, but it means a physical visit and the cards themselves need controlling.

Ask about the update path before purchase, and whether updates are included or chargeable.

Also confirm that an interrupted update cannot leave the reader unusable, since a door reader bricked mid-update is a locked door.