Procuring traffic counting: what to require, what to ask

Procuring traffic counting for a municipality means buying a traffic survey as a continuous series of vehicle, cyclist and pedestrian counts per direction and hour. The counts come from AI cameras or LiDAR sensors installed at the chosen measuring points. Bumbee Labs is the company behind the only footfall method in Europe approved by a data protection authority.

Which requirements matter when a municipality is procuring traffic counting?

Should we tender for a continuous series, or are sample counts enough?

Decide street by street, before writing the specification. A tube for two weeks answers one question once. A continuous series also shows what the roadworks did, what the new crossing changed and how summer differs, and yields AADT. Ask bidders to price both. For streets a municipality governs on trends, buying samples is the most common and most expensive mistake.

Three bids, three methods: how do we compare them?

A tender should put the same questions to every bidder and leave the technology open: which streets, which movement classes, series or samples, and how the data is delivered and validated. Each bid then shows how its method answers them, whether tubes, radar, LiDAR or AI camera counting; the automatic traffic count guide compares the methods.

What validation should we require from bidders?

A municipal tender should ask each bidder how a deployment is checked before the data is accepted: a hand count on the same street in the same hours, compared with the system’s own counts for each movement class that is bought. Bumbee Labs runs this check on every deployment (see data deliverables).

Can we keep the data if we change supplier?

Make sure the tender requires it. Write open, documented delivery into the specification: an API and agreed exports, aggregation levels, and explicit data ownership after the contract ends. A dashboard is one way to read the series; the specification should secure the series itself. Our People counting API is open and documented.

Hardware or service: who carries calibration and maintenance?

A municipality that buys sensors takes on their calibration, upkeep, data processing and replacement. Procuring the measurement as a service, such as LiDAR traffic counting run by the supplier, moves that work to the supplier, who answers for validated data at an agreed quality.

Which other questions come up in a traffic counting tender?

• Do we have to buy car, bike and pedestrian counts separately? Not if the specification requires movement classes to be separated in the same scene. One installation then delivers vehicles, cyclists and pedestrians as distinct series. One technology per class means three suppliers whose reports do not fit together. For cyclists, see bicycle counting.
• What happens when a sensor breaks? That depends on who owns the hardware. Bought sensors leave maintenance and replacement with the municipality, while under a measurement service the technical risk stays with the supplier.
• We procure through our traffic consultant. The consultant receives the same series as an API feed or as exported files, by direction and hour. Consultancies that want to work with us can read about the partner programme.
• Our town centre and our traffic are procured together. Pedestrian and vehicle counts can then arrive in one data flow, following the principle in the procurement guide for people counting: buy the data and the right to keep using it.

How do we write the data protection requirement so the data protection officer (DPO) says yes? On public streets, privacy decides what the legal team must review and what citizens accept. Write the requirement in technical terms: no imagery is retained, no identities are captured and re-identification is not possible. A supplier whose sensors cannot produce images answers this in one paragraph. See GDPR and traffic measurement.

What does a traffic counting service from Bumbee Labs include?

Bumbee Labs delivers traffic counting as a measurement service: continuous series separated by movement class, from LiDAR and AI camera installations, through an open API alongside pedestrian and footfall data. If you are writing a specification, talk to us early.