For twenty years, retail has measured its physical stores with one number: footfall. A door counter. A beam across the entrance. Every time something interrupts the beam, the number goes up.
The trouble is, that number lies.
The same shopper who walks in, walks out to grab a coffee, walks back in, walks to the changing room, and walks back to the floor counts as five "visits." The family of four who arrive together count as one. The delivery driver who steps in to drop a parcel counts the same as the customer who spent forty minutes deciding on a jacket. The beam doesn't know. The beam can't know. It was never designed to.
This is why retailers still cannot answer the most important question in their business: how many unique humans walked our store today, and what did they actually do?
The Digital Mirror Image
Online retail solved this problem a decade ago. Google Analytics doesn't count visits in any meaningful sense — it counts unique visitors, sessions per visitor, time per session, page-by-page journeys, conversion funnels, drop-off points, and return frequency. No e-commerce operator on earth would accept a "door counter" as their analytics stack.
Yet physical retail — the channel that still drives the majority of global retail revenue — runs on exactly that.
Pulses.ai exists to close that gap. We are, in the simplest possible framing, Google Analytics for the physical world.
Why the Difference Is Edge AI
Here is where Pulses separates itself from every other people-counter or footfall device on the market. Almost every competitor in this category falls into one of three architectures:
- The beam. Cheap, ubiquitous, and fundamentally incapable of distinguishing one human from another. Double-counts everything. Tells you nothing about who, what, or why.
- The cloud-bound camera. A camera in the ceiling that ships raw video — or raw frames — to a remote server, where AI inference happens hours later in a batch process. Even the ones that call themselves "AI" almost always do the heavy lifting in the cloud. That means latency. It means raw video leaving the store. It means privacy exposure your legal team should already be worried about. And it means re-identification across cameras either doesn't happen, or happens by uploading biometric data to a third party.
- The dumb IP camera with bolt-on analytics. A security camera with a software layer. The analytics quality is whatever the cloud vendor felt like building. The hardware was never designed for inference.
Pulses is none of these.
Pulses Vision is a purpose-built Precision Vision Module with an Nvidia-powered inference engine running on the device itself. Every visitor is identified — anonymously, on-device, in under 100 milliseconds — and re-identified across the entire store as they move from the entrance, to the womenswear floor, to the fitting room, to the till, and back out. No raw video is uploaded. No biometric template ever leaves the optic. The intelligence is sealed at silicon level by Nvidia's chip-level encryption — Pulses holds 1 of 40 global Nvidia master keys for this purpose.
The result: the same person is the same person across the store, even though no image of them is ever stored or shared.
What Retailers Actually Get
With Pulses CX deployed across a store, the retailer finally sees their physical business the way they see their website. Unique visitors. Real ones. Counted once per person per visit, no matter how many cameras they cross. Time-on-store and time-per-zone — the physical equivalent of session duration and page time. The conversion funnel: entered the store, reached the floor, engaged with a product, entered the fitting room, arrived at the till, purchased. Every drop-off is visible. Every chokepoint is named.
VIP recognition — strictly opt-in. For customers who have explicitly enrolled (loyalty programmes, private-client services), the store knows the moment they walk in. Everyone else is counted as an anonymous unique, never identified, never stored. Dwell-time intelligence shows where people pause, hesitate, or walk past — and return frequency, anonymised on-device, tells the store whether a unique visitor has been in before, without ever knowing who they are.
Privacy Is Not a Trade-Off
The reason a cloud-bound competitor cannot do any of this responsibly is that responsible re-identification requires keeping biometric data inside a tightly controlled boundary. Once raw video or facial templates leave the optic and hit the open internet, the privacy story collapses.
Pulses doesn't let that happen. Inference is at the edge. Encryption is at silicon. Network traffic is outbound-only. The store gets richer data than its cloud-bound competitors and a stronger privacy posture at the same time.
"Insights, not images." The retailers who win the next decade will be the ones who finally measured their physical floors with the same rigour they've applied to their websites for twenty years.
Ready to begin?
Want to see your store the way the internet sees a website? Request a free audit at pulses.ai.
