Where occupants go next: how often an occupant's visit to one of your buildings was followed by a visit to another, over a range of dates.

A journey is one visit followed by the same occupant's next visit, when the next visit is to a different building and starts no more than max_gap_minutes after the first one ends (240 by default). The limit keeps a journey to one stretch of activity, so a visit on Friday evening followed by one on Monday morning isn't a journey. Visits are paired in the order they started, and only visits on the days of the week you ask for are paired: with days_of_week set to Monday to Friday, a Friday visit and the following Monday's are consecutive, and too far apart to count.

Journeys cover the buildings your credentials can access, so there is no building parameter. A journey to or from any other building isn't counted.

group_by chooses how journeys are grouped:

• route: one row per route, an ordered pair of buildings, most journeys first. Library → Gym and Gym → Library are different routes.
• building: one row per building, with the journeys that start or end there and connected_buildings, the number of other buildings it shares a journey with in either direction. Buildings are ranked by connected_buildings, then by journeys, then by building ID, so journey_count doesn't always decrease down the list. Each journey counts at both of its buildings, so these shares add up to 2, not 1.
• hour: one row per hour, 00 to 23, empty hours included. A journey is counted in the hour its first visit started, in the local time of that visit's building. With then_by=route, each hour also lists its routes in children, each with its share_within the hour.

top_n limits the rows returned: routes or buildings, or with group_by=hour and then_by=route, the routes in each hour. It's 100 when omitted. There's no maximum: for every row, ask for a top_n at least as large as the number of rows, which coverage gives (and each hour's truncation, for the routes in an hour). For a large organization, every row can make a very large response. All 24 hours are always returned. Shares are always out of every journey, including those past the limit. With group_by=hour, each hour's truncation says what the limit left out; for a route's share of only the routes returned, divide its journey_count by truncation.journeys_returned. Routes with the same number of journeys are ordered by origin building ID, then destination building ID.

A building without a name is returned with a null name and keeps its place in the order. Building IDs are always included.

start and end are dates in each building's local time, and both are included. The range must cover at least two days and at most two years (731 days). days_of_week limits the visits paired to those weekdays, and the range must include at least one of them.

The response has three parts:

• data: the rows described above.
• applied: the parameters used, including any defaults you left out.
• coverage: n_journeys, the number of journeys counted, and depending on group_by, how many routes, buildings or buildings in scope that covers.

Figures for the most recent days can change from one day to the next as more activity is taken into account. Figures for older dates are stable, so allow for this when comparing two requests that cover recent dates.

HTTP compression is enabled for responses larger than 500 bytes when the request includes "gzip" in the Accept-Encoding header.

The journeys endpoint is rate-limited to 60 requests per minute.