People Counters are DEAD. Here’s why AI is taking over Retail Analytics.

When people-counting sensors launched 10+ years ago, they were great. For the first time, retailers could measure foot traffic and understand store visits.

But like any tech that fails to adapt, they’ve become outdated.

After speaking with 100+ retailers, here are the top 5 reasons they’re walking away from people-counting sensors:

1. Accuracy isn’t what it seems.

That 99% accuracy? Only in empty stores with perfectly spaced customers. Real stores are crowded, chaotic, and full of movement. These sensors detect shapes, not individuals. And when your business relies on real data, guesses just don’t cut it.

2. They only tell you someone walked in. Nothing else.

Did the customer explore the store? Look at a product? Leave because of long lines? Traditional sensors can’t answer those questions — and that context is what actually matters.

3. When they fail, you won’t notice. But your data will.

Most sensors aren’t actively monitored. And no monitoring means failed sensors keep feeding bad data. 18% fail in the first year. And if you catch it, fixing it means a tech with a ladder in your store during peak hours. Good luck with that.

4. They don’t scale. Because of cost.

Rolling out hardware across hundreds of stores adds up fast — sensors, installation, maintenance. Multiply that by your entire store network and suddenly you’re spending more on infrastructure than on insights.

5. Want new features? Buy new hardware. Because your current setup? It’s obsolete.

That’s the business model. You’re locked into a system where progress means replacing equipment. Like old phones that stop supporting new software, your sensors become useless the moment an upgrade drops. Staying current means starting over.

Here’s what’s next:

AI and Computer Vision.

No more sensors.

Just better insights using the cameras you already have.

Full visibility of in-store customer journeys — with none of the outdated tools.

Retail is evolving. The way we measure it should too.