vibe
Get the app
A lab that only shows its conclusions is a marketing department. Where we can share the method and the data, we do — including the parts that did not work.
Where we can share the method and the data, we do. Where a result rests on a partner's commercial data, we publish the method and withhold the rows.
Matched venue pairs across event and non-event nights. The closest pair gained 1,793 movements per event; leakage to the next precinct ran 1.8× on stadium nights.
Why the capacity and opening-hours fields on most venue APIs are seed defaults, how to detect it, and what it does to a ranking model.
Two phones in the same room is a harder fact than two accounts following each other. What that unlocks, and what it must never be used for.
A venue joins the network. Presence is sampled continuously, not at a door, and the raw series stays tied to a place rather than a person.
A venue is only legible against its own ordinary week. We model a per-venue, per-weekday, per-hour expectation before we call anything unusual.
Event nights are compared against matched non-event nights at the same venue — same weekday, same season, similar weather — not against a city average.
Every figure ships with what it rests on and what would falsify it. If the confidence is low, the number is labelled low, not rounded up.
The measurement does not need to know who you are, so it does not. This is a design constraint in the hardware, not a policy we could quietly relax later.
If you are studying night-time economies, planning late transport, or arguing a licensing case, the measurement is more useful in your hands than ours. We share aggregate data with research and public bodies.
Aggregate, precinct-level series for published research. We ask for a method note and a citation, not a fee.
Councils and transport authorities planning around the night-time economy. Tell us the decision you are trying to make.
Venue groups, stadiums and precincts. This runs through the measurement product rather than the lab.