Crowd-Sense

Density measurement that ships counts, not identities

A passive Wi-Fi pedestrian sensor with on-device aggregation. Device identifiers never leave the Collector; only aggregate counts do. Designed for the safe-harbour described in the ICO's 2016 Wi-Fi Location Analytics guidance, and built for councils, BIDs, retailers and workplaces who need defensible footfall data in 2026.

Or read the DPO Technical Note (PDF), the artefact a council DPO reads first.

Crowd-Sense data flow: packets in, aggregate, counts out Plan view. Anonymous figures stand in three RSSI distance brackets around a Collector. Each emits inward signals representing Wi-Fi probe requests. The Collector aggregates them into bracket counts, discarding identifiers. Only counts leave, going to a Reporting Portal. A B C COLLECTOR A: 0 B: 0 C: 0 identifiers discarded; only counts leave { A:12 B:27 C:41 } aggregate counts REPORTING PORTAL

Why Crowd-Sense

Most Wi-Fi analytics products from the past decade hash MAC addresses and retain them in a back-end database. Crowd-Sense doesn't. Identifiers are discarded on the Collector at aggregation close. There is no off-device dataset to re-balance, subpoena, or breach.

01 / Privacy

Device identifiers never leave the Collector.

Not hashed, not salted: discarded at each aggregation-window close. The wire payload contains no identifier of any kind.

02 / Methodology

Honest by default.

We publish the MAC randomisation overcount curves. Density and trend are reliable; absolute unique-visitor counts at long windows are not, and we say so.

03 / Provenance

Established 1994.

Visual Solutions (UK) Ltd has been operating continuously since 1994. Procurement-grade due-diligence, not a fresh-vehicle startup.

The curve we publish.

Modern phones rotate the addresses they broadcast every few minutes (a deliberate Apple/Google privacy feature). A Collector that respects that rotation, rather than trying to defeat it, systematically overcounts unique devices at longer windows. We publish the inflation factor for every window we report. Density and trend are reliable; the absolute number is never overclaimed.

MAC randomisation overcount, by aggregation window Overcount factor rises predictably from approximately 1.6 times at 1 minute to 7 times at 60 minutes. Range band shows the full published low and high values. 10× 1m 5m 15m 30m 60m AGGREGATION WINDOW TYPICAL OVERCOUNT FACTOR
The amber band shows the full published low and high; the line is the typical observed midpoint. Read the full methodology and the cite-able paragraph →

The Oxfordshire moment

In January 2026 Oxfordshire County Council publicly withdrew the November 2025 Oxford congestion-charge footfall figures after a mid-2025 dataset rebalance from its supplier produced anomalies the council couldn't defend.

The lesson generalises. Any organisation that buys footfall data without understanding the methodology behind the numbers is one supplier-side change away from the same press cycle. Crowd-Sense is built to give a different answer: each Collector produces counts directly from its own local observations. There is no panel to rebalance, no third-party SDK, nothing that could change without us deliberately changing it.

What we mean by "no panel": we don't aggregate location data from mobile apps. We don't buy data from brokers. Each Collector passively listens for nearby Wi-Fi probe requests, counts them on-device, and emits aggregate bracket counts. That's the whole pipeline.

Read our published methodology and accuracy disclosure →

Talk to us

Free 60-day pilot for qualifying organisations. Self-install in minutes. Data live within the hour.

Book a 30-minute briefing