The Open Office Is Made Private by Its Noise Floor, Not Its Silence

The complaint that follows everyone back to the office

Across workplace surveys from 2024 to 2026, one grievance keeps rising to the top: noise. Leesman, which has polled more than 350,000 employees across thousands of workplaces, finds that roughly three quarters of people call managed noise important, while only about a third are satisfied with what they actually hear. Dissatisfaction with sound turns out to be the single strongest predictor of a workplace that feels unproductive, stronger than temperature, stronger than light. As companies press for a return to the office, acoustic privacy has become one of the deciding factors in whether the commute feels worth it.

The instinct is to fight noise by removing it: thicker carpet, more ceiling baffles, taller felt screens between desks. That instinct is half right and half a trap.

Privacy is a ratio, not a volume

The thing that actually disturbs a person at a desk is not loudness. It is intelligibility, the fact that a neighbour's phone call arrives as words the brain cannot help decoding. Overhearing meaning, not merely sound, is what breaks concentration and erases the sense of privacy.

Whether speech is intelligible at a given seat is governed by a ratio: the level of the intruding voice against the level of the steady background sound already present. Acousticians measure the result as the Speech Transmission Index, STI, running from 0 (nothing understood) to 1 (every word clear), and as the older Articulation Index. Offices feel private when intruding speech drops to an Articulation Index around 0.20 or below.

The open-plan version of this is codified in ISO 3382-3, which describes a room by how fast speech fades as you walk away from the talker. Its headline number is the distraction distance: the point at which STI falls through 0.5, below which a voice stops being clearly followable. A good open plan pulls that distance inside about 5 metres; a poor one lets it stretch past 10, so a single conversation colonises a whole floor.

Why absorption alone cannot win

Here is the trap in the removal instinct. Absorption, carpet, ceiling tiles, baffles, lowers reverberation and takes the hard edge off a room, and it genuinely helps. But it lowers the intruding speech and the background hum together. The ratio between them barely moves, and intelligibility can stay exactly where it was, or even rise, because a quieter room lets a distant voice ride cleaner and more decodable above a now-hushed floor. A library-quiet open office is often the worst kind: every whisper carries.

Barriers have the same limit. A screen has to break the line of sight and rise well above seated head height to shadow speech, and even then sound diffracts over the top and reflects off the ceiling around it. Raising screens hits diminishing returns fast, and the taller they get the more they wall people into the very isolation the open plan was meant to avoid.

That is why acoustic design is described as A-B-C, in order: Absorb, Block, and then Cover. The first two shape the room. The third supplies the piece the room cannot make for itself, a controlled background level that sets the denominator of the ratio.

Covering: an engineered floor, not white noise

Covering is sound masking: a specifically engineered background sound, played through a grid of small ceiling emitters, tuned to sit in the same frequency band as human speech, roughly 250 Hz to 4 kHz. It is not the hiss of white noise, which sounds like an air leak and skews bright. The modern target is a shaped spectrum, close to filtered pink noise with a gentle downward slope of around 5 dB per octave, engineered to smother speech frequencies while reading to the ear as bland, ignorable air movement.

The level matters more than anything. Industry practice sets open-plan masking to about 45 dBA, commonly a window of 42 to 48, and holds it uniform within a tight tolerance, on the order of half a decibel from seat to seat. Too low and it does nothing; too high and it becomes its own distraction. After install, a technician walks the floor with a meter at ear height, tuning each small zone to the target curve, because a masking level that swings from desk to desk is as disturbing as a leaky faucet you can never quite locate.

Raise that floor a few decibels and the arithmetic shifts. The intruding voice is unchanged, but the denominator has grown, the ratio falls, STI drops, and the distraction distance contracts toward the talker. Speech dissolves into an unintelligible murmur a few desks away, present enough to know the room is alive, blurred enough that the words no longer hijack attention.

The floor that follows the room

The frontier is making that floor adaptive. A fixed masking level is a compromise: loud enough for the busy afternoon can feel heavy at 8 a.m. in an empty wing. Level-adaptive systems, studied recently at Eindhoven University of Technology, sense the ambient level and nudge the masking up and down with the room, keeping the ratio in the right band through the day. Field work on such systems points to lower reported distraction and better coping with noise, without the fatigue of a wall of sound held at full volume from dawn.

The lesson survives every fashion in office layout. You do not make an open plan private by chasing silence. You make it private by controlling the ratio between the voice you should not hear and the engineered floor beneath it, and that floor is something you add, not subtract.