Where they went

Billboard passes and traffic flow

Every time one of our camera vehicles passes a roadside billboard on a map-matched route: seconds in view, seconds legible, whether the cab was occupied, how many passengers, and the camera's attention readings, summed per board per local day. The flow table adds modelled vehicles per hour past each face by day type and hour, from US federal and Quebec traffic counts shaped by our fleet's own trips.

Request a sample All datasets

The question it answers

How many people passed this billboard, could they read it, and who was in the car?

24,049

billboards passed

15.7 million

passes

media owners

139

DMAs the boards stand in

Measured over the 90 days to September 25, 2026.

The dataset

Grain: Daily.
Cadence: Daily rows per board, one file per day. The flow table is one current snapshot.
History: From May 12, 2026.
Format: Parquet or JSONL.
Delivered to: Your S3 bucket, your GCS bucket, or a time-limited download link, with a manifest and a schema.
Method: How each measure is produced and how to add rows up: the measurement methodology.

What buyers do with it

Audit an out-of-home buy against measured passes.
Rank faces by legible seconds and occupied passes.
Price boards on who was in the car.

The tables

Read from the export registry each time this site is built.

`board_pass_daily`

Daily, from May 12, 2026

Grain: One row per local_date and board_id.
Delivery: One file per day, as Parquet or JSONL.
Basis: Measured.
Definition: Version 1.

Column	Type	What it holds
`local_date`	date	The board's local calendar day.
`board_id`	string	The roadside face: the venue id of the board face.
`venue_type_id`	integer	OpenOOH venue type id of the face (0 = unclassified).
`publisher_name`	string	Media owner of the face (empty when none is recorded).
`dma_code`	string	Nielsen DMA of the face. Null: The face is outside a DMA.
`passes`	integer	Passes of our map-matched vehicles with the face in view. Unit: passes.
`vehicles`	integer	Distinct vehicles behind the passes. Unit: vehicles.
`site_primary_passes`	integer	Passes on which this face was the site's primary face. Unit: passes. Null: The currency rollup holds no row for the board-day.
`expected_passes`	double	Face-share-weighted passes, for a face whose orientation is unresolved. Unit: passes. Null: The currency rollup holds no row for the board-day.
`visible_s`	double	Seconds the face was in view over the passes. Unit: seconds. Null: The currency rollup holds no row for the board-day.
`legible_s`	double	Seconds the face was legible (available to be read) over the passes. Unit: seconds. Null: The currency rollup holds no row for the board-day.
`occupancy_known_passes`	integer	Passes whose cab occupancy is known. Unit: passes.
`occupied_passes`	integer	Of those, the passes with a passenger in the cab. Unit: passes.
`occupied_share`	double	occupied_passes / occupancy_known_passes. Unit: share. Null: No pass with a known occupancy (occupancy_known_passes = 0).
`pax_count_sum`	double	Passengers summed over the occupied passes that carry a count. Unit: passengers.
`pax_count_n`	integer	Occupied passes that carry a passenger count. Unit: passes.
`pax_count_mean`	double	Passengers per occupied pass: pax_count_sum / pax_count_n. Unit: passengers_per_pass. Null: No occupied pass carries a passenger count (pax_count_n = 0).
`camera_windows`	integer	Cabin camera windows during the passes. Unit: windows.
`hy_pose_valid_windows`	integer	Head-yaw instrument: windows with a valid pose. Unit: windows.
`hy_heads_up_windows`	integer	Head-yaw instrument: windows with the head up. Unit: windows.
`hy_eyes_out_windows`	integer	Head-yaw instrument: windows with the eyes out of the vehicle. Unit: windows.
`sh_pose_valid_windows`	integer	Shoulder (MoveNet) instrument: windows with a valid pose. Unit: windows.
`sh_heads_up_windows`	integer	Shoulder instrument: windows with the head up. Unit: windows.
`sh_eyes_out_windows`	integer	Shoulder instrument: windows with the eyes out of the vehicle. Unit: windows.
`pass_audience_method_version`	integer	pass_audience method version of the audience rollup. Null: The board-day was built under more than one method version.
`exposure_currency_method_version`	integer	exposure_currency method version of the currency rollup. Null: More than one method version, or no currency row.

Also answered live by the measurement API: Who was in the cab during a board pass.

`board_flow_hourly`

Hourly, from Current state

Grain: One row per board_id, dow_type and local_hour.
Delivery: One file per delivery, as Parquet or JSONL.
Basis: Modelled.
Definition: Version 1.

Column	Type	What it holds
`board_id`	string	The roadside face: the venue id of the board face.
`dow_type`	string	Day type: weekday, saturday or sunday (holidays count as sunday).
`local_hour`	integer	Board-local hour of the day, 0-23.
`flow_vph`	double	Modelled vehicles per hour past the face. Unit: vehicles_per_hour.
`flow_cv`	double	Coefficient of variation of flow_vph. Unit: ratio.
`aadt_facing`	double	Annual average daily traffic facing the face, after the two-way split. Unit: vehicles_per_day.
`ways`	integer	Road ways contributing to the face. Unit: ways.
`aadt_source`	string	Weakest AADT basis among the ways: hpms (FHWA HPMS), mtq (Quebec) or imputed.
`aadt_year`	integer	Oldest AADT year among the ways. Null: The AADT is imputed.
`profile_basis`	string	Basis of the hourly shape: fleet_blend (our fleet's observed traversal profile blended with the TMAS prior) or tmas_prior (the prior alone).
`kappa`	double	Strength of the prior against the fleet profile in the blend. Null: The build used the TMAS prior alone.
`fit_window_end`	date	Last day of the fleet profile's fit window.
`exposure_currency_method_version`	integer	exposure_currency method version of the flow.

Request a sample

Tell us the markets and the weeks you want. The sample is cut from production by the same export that makes the full delivery.

Request a sample

All datasets · Next: Radio presence and device exposure