How we calculate these numbers

A passage is the raw transit of a person within the screen's field of view. The source varies by location: counting sensors on the device itself, data provided by the space operator (metro, stations, car parks) or, where unavailable, a conservative estimate based on comparable nodes.

The figure is a daily average per location. Reliability is highest where there is proprietary sensing, medium where the data comes from the operator, indicative where it is an estimate: the source applied to each location is stated in the contact proposal.

An impression, or opportunity-to-see, is the number of passages that translate into an actual exposure to one of the advertiser's slots. It is the footfall multiplied by two factors: the exposure factor (share of transits facing the screen while the ad is on) and the share of the loop assigned to the advertiser.

The exposure factor depends on the geometry of the node (corridor, platform, concourse) and is measured or estimated per location. The loop share equals 1 divided by the number of slots in the loop: with a loop of 6 ten-second slots, each advertiser occupies one sixth of screen time.

Net reach is the number of distinct individuals reached at least once in the period. It does not grow linearly with duration: at a transit node the same person passes dozens of times a month, so beyond a certain point adding days increases frequency, not coverage.

We model reach with a saturation curve that tends towards a ceiling — the pool of addressable individuals at the node: net reach = pool × (1 − e^(−(exposed passages × k × days) / pool)), where k is the shape constant of the curve. Doubling the duration grows reach by less than double.

Average frequency is the ratio impressions / net reach: how many times on average the same person sees the message. On transit inventory a high frequency is a planning strength, not something to hide.

With multiple locations the reach of the individual nodes is summed and reduced by an inter-node overlap share (individuals who visit more than one node), configurable and zero by default.

Screens run during the opening hours of the node that hosts them. The loop share per advertiser and the slot duration are declared per location; the default values are one ten-second slot in a loop of six slots.

While the campaign is on air, Bufaga devices at the same locations capture particulate matter. The primary KPI is PM removed (PM10 + PM2.5) in grams, calculated as capture capacity per device × number of devices × campaign days.

The per-device capture capacity, when measured in the field, is the declared one; without a measurement a conservative estimate is used: 2,000 m³/h airflow × 20 µg/m³ average urban PM × 80% filtration efficiency ≈ 0.768 g of PM per device per day.

Equivalent trees and offset car kilometres are CO₂ equivalences, not PM: they derive from the CO₂ captured by the device over the period. A mature tree absorbs about 25 kg of CO₂ per year (FAO / urban forestry); an average petrol car emits about 120 g of CO₂ per km (EU fleet target reference).

The public impact calculator uses the same PM model in a simplified form: PM removed = 0.768 g/day × the hypothetical number of devices × days, where days are derived from months using the convention 1 month = 30 days. It shows neither trees nor car kilometres, because those equivalences depend on a CO₂ model that is not yet consolidated.

The calculator's "clean breaths" are the volume of air passed through the devices expressed in breath-equivalents: airflow 2,000 m³/h × 24 × days ÷ the average resting breath volume (0.5 litres, ~7 ml/kg). It is a treated-volume figure, with no efficiency multiplier applied.

The per-city context line compares the result with the selected city's average annual ambient PM10 (orders of magnitude from EEA / regional ARPA urban station annual means; dataset and reference year being verified).