The device that is not transmitting

If you only ask a prospective provider one technical question, ask this one: how do you handle a device that is not transmitting? It requires no expertise to ask, the answer is highly diagnostic, and it goes directly to the assumption that most weak engagements are quietly built on.

Why it is the right question

The popular mental model of a bug sweep is someone walking around with a receiver, looking for a signal. That model is not wrong, it is incomplete, and the gap is large enough to drive most of the industry's failures through.

A spectrum analyser detects energy that is present while you are listening. It follows that anything not radiating at that moment is invisible to it. That covers several entirely ordinary categories of device:

• Recorders that store locally and are collected later. They never transmit at all, so there is nothing to detect by radio at any point.
• Devices that are remotely switched, dormant until someone activates them — often precisely when something worth hearing is scheduled.
• Burst transmitters that store and then send in a brief packet at intervals, potentially hours apart. Your sweep has to coincide with the burst.
• Devices operating on a schedule, active only during working hours — which is an interesting interaction with the sensible practice of sweeping out of hours.
• Frequency-hopping devices, which are detectable but require an operator who knows how to look, rather than one waiting for an obvious peak.

None of these are exotic. They are common properties of commercially available hardware, and any of them defeats a purely RF-led engagement completely.

What a competent answer covers

The answer you want describes layering, because the whole discipline is layering. Different methods have different blind spots, and the engagement is designed so that the blind spots do not line up.

Non-linear junction detection is the direct response: it finds semiconductor junctions whether or not the device is powered, which means a dormant recorder in a wall cavity is in scope. Thermal imaging catches powered devices generating heat inside fabric and fittings. Physical search — systematic, fingertip-level, unglamorous — is in my view the most consistently productive of them, and it does not care whether anything is switched on. Extended or repeat monitoring becomes proportionate where the environment justifies it and the burst-interval problem is real.

A good provider will also tell you the limits of those, unprompted. Non-linear junction detection responds to semiconductors generally, so a building full of electronics generates a great deal to eliminate, and the skill is in the elimination. Thermal imaging needs a temperature differential. Physical search is bounded by time and by what can be opened without damage.

What a bad answer sounds like

There are two failure modes and they are easy to recognise. The first is not understanding the question — an answer that restates the RF process in more detail, as though the issue were sensitivity rather than the device being silent.

The second is more confident and more concerning: a claim that their equipment finds everything. Nothing finds everything. A provider who says otherwise either does not know the limits of their own instruments or is willing to misrepresent them, and neither is something you want in your boardroom ceiling.

The instrument is not the competence. A spectrum analyser in untrained hands produces confident nonsense, and confident nonsense is exactly what a clean report looks like.

This is why I published the method rather than describing capability in the abstract. Method can be checked, compared and argued with. Capability claims can only be believed or disbelieved, and asking a buyer to simply believe you is not a standard.