This page explains where Selectivv data comes from, how we process it and how we check its quality. Every report comes with a project-specific methodology note: period, area, definitions and group sizes.
Where the data comes from
The source is anonymised mobile advertising identifiers (MAID: Google Advertising ID and Apple IDFA) visible in the advertising ecosystem and RTB. When an app or mobile website shows an ad, it sends an ad request with the device identifier, context and, if the user allowed it, location.
- Scale: 350,000+ apps and 17M+ mobile websites as signal sources, 20M+ behavioural profiles in Poland.
- Coverage: data is available globally, except Russia, China and North Korea. We have delivered projects in over 20 countries.
- What we see: in which apps and on which websites an ad appeared, and where the device was. We do not see what someone does inside an app or the content of messages, and we do not work with personal data.
How we link online and offline
Each anonymous device profile combines two types of information:
- Online: categories of apps and websites the device uses, grouped into interests and purchase intent, e.g. finance, automotive, gaming.
- Offline: places visited, matched to a database of points of interest (POI) in 44 categories, e.g. shopping centres, petrol stations, gyms, and to areas such as municipalities, districts and the H3 hexagonal grid.
A visit is recognised when a device is present within a defined area (geofence). Geotrapping® accuracy is 2-4 m. Geotrapping® is Selectivv’s proprietary method for reaching people who have been to a place before.
How fresh the data is
- Daily data since 2015. You can compare seasons, years, and periods before and after an investment, store opening or campaign.
- The OmniFonia dashboard receives new data every month.
- Custom studies cover the period you choose, including historical periods such as last tourist season.
How we ensure quality and representativeness
Mobile data is not a random sample but a very large set of observations. That is why we ensure quality at several stages:
- Data cleaning. We discard signals with unreliable location and devices with unnatural activity, such as automated traffic or impossible movements.
- Sample size. Analyses are based on large groups of devices. For small areas or short periods we choose the analysis period so the group is large enough, and we flag estimates in the report.
- Reference data. We compare results with public statistics, such as Statistics Poland (GUS), and with client data such as footfall counters, sales or tickets. This lets us convert device counts into estimated numbers of people and catch deviations. Example: in Łódź, mobile data showed 170,000 more residents than official statistics, because we count presence, not registration.
- Control groups. In campaign measurement we compare people or regions exposed to an ad with similar ones that were not, and with the period before the campaign. This separates the campaign effect from seasonality, weather or promotions.
- Aggregated results only. We report groups and areas, never individual devices or people.
Limitations worth knowing
- The data covers devices visible in the advertising ecosystem. Groups that use smartphones less, such as young children and some seniors, are under-represented. We account for this in interpretation and when estimating numbers of people.
- One person may have more than one device, and a device may reset its advertising ID. Results about numbers of people are therefore estimates, not a census.
- Location accuracy depends on the device signal. In dense areas and multi-level buildings we define areas with an appropriate margin.
Privacy and GDPR
We work with anonymised advertising identifiers, not personal data. We do not use third-party cookies. Results are delivered only in aggregate, for groups and areas, and we never share data that could identify individuals. More in our privacy policy.
What you get with the results
- a project methodology note: source, period, areas and definitions (e.g. what counts as a visit, resident or tourist),
- the size of the groups behind the results,
- which values are estimates and within what range,
- on request: a comparison with reference data you provide.
Have a question about the methodology of a specific study? Contact us and the analyst leading the project will answer.