Methodology

Where the rows come from

Every row traces back to one of three source kinds. Brands' own store locators, extracted directly rather than taken from a third-party directory. Government registries, such as insured bank branches, certified hospitals, schools, and airports, each carrying its own identifier. Open geospatial data, from Foursquare Open Source Places and Overture Maps, used to fill the category and geographic coverage the first two do not reach.

If a location is not in a brand's own locator, a government registry, or an open dataset, it is not in the dataset. There is no device location data in it, no carrier or app panel, and no location-history modelling.

Fusion and dedup

The same physical location often appears in more than one source: a brand's own locator and an open dataset, or two overlapping registries. Records are matched on name, address, and geographic proximity, then conflated into a single canonical business location when they describe the same place.

The surviving record's fields are chosen by source priority. A brand's own locator outranks an open-data supplement for that brand's own locations. Every canonical location carries a stable, content-derived id and, where one exists, the original source's identifier, so a row can always be traced back to what produced it.

Confidence is scored, not hidden

Field-level confidence follows source agreement. A name, category, or address corroborated across independent sources scores high. A single-source field, or one with conflicting values across sources, scores lower.

Confidence drives internal quality sampling and category assignment. It does not drop or hide rows. The per-field fill rates published on each brand page are the customer-facing view of the same signal.

Quality gates

Every monthly release runs a fixed set of automated tests and release checks before it publishes. They cover the schema contract, per-tier row-count sanity, address and category field validity, id-format compliance, and duplicate detection. A release that fails a gate does not ship, and the store always serves the last release that passed all of them.

Release cadence and change events

Most brand locators are refreshed weekly, so a single brand can be more current than the dataset's own publish cycle. The full dataset is reassembled, conflated, and published once a month. That monthly release is what every product page, price, and sample reads from, so the whole dataset stays point-in-time consistent even though sources refresh on their own schedules.

Change between releases is reported as distinct events: openings, closures, relocations, renames, corrections, and source-only changes. Net location change reconciles to the event totals. Every published metric carries a reproducible query and a release identifier.

What ships with a download

Every download ships with its own NOTICE file, and that file reflects what is actually in the archive. Downloads that include Foursquare-derived or Overture rows carry the upstream open-data license texts alongside it. Government-registry sources are public records, and where an agency publishes its own attribution requirement, that credit travels with the rows.

Ground Truth US carries no Foursquare or Overture attribution, because neither source is in it. The NOTICE is fixed per release, not per purchase.

Known limitations

Brand coverage grows every month but is not exhaustive. Smaller and regional chains are still being onboarded. Independent, non-chain businesses are covered only where a registry or open dataset includes them, so the dataset is strongest on chains and registered institutions.

There is a detection lag between a real-world opening or closing and its appearance in the change feed, bounded by that brand's refresh cadence. Foursquare-derived and Overture rows carry lighter per-field verification than the brand-extracted and gov-registry tiers. That is what the tier system is for: it tells you which guarantee applies to which row, instead of asserting one confidence level over the whole file.

What is measured

Out-of-home advertising runs in the physical world, so we measure it there. Mogean matches advertising screen interactions against mobile device observations to identify the devices that could see a campaign. It then follows those devices to the advertiser's locations.

Devices are matched to screens on proximity, trajectory, velocity, and timing. Devices are matched to destinations on the same signals plus dwell time, which separates visitors from passers-by.

Exposure is resolved per screen

Exposure is resolved per screen, per creative, per venue, and per timestamp. Every device carries that context through to delivery, so a result can be read down to the individual screen rather than only at market level.

The same pipeline runs one screen and one destination, or tens of thousands of screens and thousands of venues. The method does not change with the size of the campaign.

How the signals are handled

Mogean receives these signals under separate agreements covering their source, their permitted uses, and the opt-outs that must be honoured. Mogean does not collect them from this website.

Inside the processing pipeline the signals are hashed. Foot traffic reports are delivered as aggregate results, not as device lists. Where an exposed audience is delivered, it goes to the client who commissioned it, under a separate agreement governing what that client may do with it.

Signals and identifiers are kept for as long as we need them to produce and support the measurement a client commissioned, and are then removed from the working data.