Data Sources

Every screen is evaluated from its own mix of data sources. Which sources apply depends on where a screen is and what surrounds it -- a roadside billboard, a screen in a pedestrian zone, and an indoor screen without public access are evaluated on fundamentally different data.

How sources are selected

There is no single data feed behind Adlocaite. For each screen, the platform selects the sources that actually describe its audience, based on:

Regional availability

Geographic location and local data availability
Regional data partnerships
Local regulations and data access permissions

Environmental context

Venue type (roadside, pedestrian zone, transport hub, closed venue)
Audience behavior patterns specific to the location type
Which measurement methods are physically possible at the location

The number of sources per screen therefore varies: a screen in a well-measured urban environment draws on more independent sources than one in a location where fewer data sets exist.

Data source categories

We work with the following categories of sources. Named providers are examples ("among others") -- the concrete mix can differ by region and evolve over time.

Targeting data

These sources determine a screen's targeting attributes -- who is around it, and when:

Geodata: Points of Interest and Points of Sale

Commercial directories and geographic databases of establishments around the screen, among others OpenStreetMap
POI categorization (airport, train station, university, business district)
POS categorization (fashion, electronics, supermarket, restaurant)

Official statistics

Demographic data from national statistical offices (for example Statistisches Bundesamt, Statistik Austria, CBS)
Age structure, income distribution, and population characteristics of the catchment area

Weather and calendar data

Weather forecasts for weather-based targeting, provided by The World Clone, the dbpg data platform
Public holiday calendars for holiday-based targeting

Audience measurement data

These sources feed the impression values -- the number of contacts a playout reaches -- and their development over time:

On-site frequency measurements

Footfall and passer-by measurements at the screen location
Manual counts as reference points

Movement data

Anonymized, aggregated movement flows around the location
Traffic data for road-based locations, among others from Google and TomTom

Transport movement data

Flight and train movements for screens at transport hubs, derived from public timetables and utilization patterns

Satellite data analysis

Automated analysis of satellite imagery, for example object counts (such as vehicles) as historical reference data
Imagery sourced among others via UP42 and the EU Copernicus programme

Three screens, three data mixes

The mechanism is the same for every screen: count data tells us how many people pass, movement data tells us where they come from and where they are heading. From origin and destination we derive an intent toward the POIs and POS in the area, which serves as the attribution basis. Combining intent with the counts yields an estimate of the audience in front of the screen -- not just how many people pass, but who plausibly passes, and why. The result is an expected impression value and audience profile per playout for every hour of the week, which is exactly what the matching uses when deciding whether a spot on a screen fits a campaign.

What differs per screen is the leading count signal and how far the catchment reaches. Three examples:

A roadside screen on an arterial road The count signal is traffic: vehicles per hour on this road segment, with satellite-based vehicle counts as historical reference points. The catchment is wide -- commuter flows span the whole corridor, so origins and destinations lie kilometers apart. The intent derives from what lies along the route: the retail park and the supermarket up ahead give passers-by a plausible shopping intent, the business district at the end of the corridor a commuting one. Morning and evening audiences on the same screen differ accordingly.

A screen in a pedestrian zone The count signal is footfall: on-site frequency measurements at the location. The catchment is local -- movements happen on foot between the surrounding shops and restaurants. The intent derives from that immediate POS mix: someone passing the screen is plausibly there to shop or dine, and the audience profile follows opening hours and weather far more directly than on a road.

A screen in a closed venue without public access -- for example an airport gate Behind an access restriction there are no public passers-by, so street-level count and movement data do not describe the audience. Here the venue context itself is the data source. At an airport gate, that context is the boarding schedule: passenger volumes derive from flight movements, and the audience follows the origins and destinations of the flights the gate serves -- a Monday-morning business route and a holiday charter show measurably different audiences at the same hour. In an office lobby, the context is access and usage instead: a known weekday rhythm and a professional audience by definition. In both cases the audience profile follows from the venue, not from a catchment.

Dynamic updating

There is no fixed refresh cycle -- update frequencies range from seconds to days, depending on the source and on how quickly its signal actually varies. Transport movements such as train departures change within seconds to minutes; weather forecasts within hours; demographic structures shift over months and are refreshed accordingly rarely.

The update rotation is decided by our analysis AI, which combines algorithmic rules with its own assessment per source and screen: how volatile the signal is, whether the environment has shifted, and whether a source's quality or availability has changed. This keeps values current where currency matters, and avoids unnecessary computation where nothing changes.

Screen evaluation levels

For impression evaluation, every screen is classified into one of four levels. The level a screen reaches depends on the data available at its location -- richer data enables a higher level.

Level 1

The screen has a fixed impression value per playout -- the average number of contacts at the location, without intraday variation.

Level 2.0

The screen's impression value changes over the course of the day according to a daily curve that is constant across days.

Level 2.1

In addition to Level 2.0, the daily curve differs by weekday -- a Saturday afternoon is evaluated differently from a Tuesday afternoon.

Level 3

The impression value is forecast by time-series models built on historical measurements and verified estimates. Multiple sources are triangulated into the historical foundation -- depending on the location, this includes on-site counts, movement data, transport movement data, and satellite-based object counts.

A Level 3 screen is therefore evaluated on the expected audience for the concrete playout moment, not on an average.