Real-Time People Tracking: How to Choose a Solution and a Provider

Most real-time location projects do not fail at the demonstration. They fail eighteen months later, when the site wants to add a second building and discovers the licence model punishes them for it, or when the safety team asks for a report the platform cannot produce.

Demonstrations are designed to succeed. A pre-mapped room, a handful of tags, a laptop. What you are actually buying is the awkward part that comes after: installation across a live site, integration with systems that predate the internet, and a support relationship that has to work at two in the morning.

Here is what to ask before that becomes your problem.

On accuracy

“What accuracy, at what anchor density, in what environment?” An accuracy figure without those three qualifiers is marketing. Ask for the number in a metal-racked warehouse, not an empty office.

“Show me the error distribution, not the average.” An average of one metre with a long tail of eight-metre errors will generate false alerts until your team stops trusting the system. The ninety-fifth percentile is the honest number.

“How does it behave when an anchor fails?” Graceful degradation to a wider zone is acceptable. Silence is not.

On the commercial model

“Price me at three sites and four thousand tags.” Per-tag pricing that looks reasonable at pilot scale becomes the entire objection at rollout. Model it early.

“What is the tag replacement cycle and who pays for it?” Battery life claims assume a beacon interval. Ask what interval, and what it does to accuracy if you slow it down to save batteries.

“Is there a charge to add a floor plan, a zone, or a user?” These are the line items that surprise people in year two.

On data and integration

“Who owns the location data, and can I export the raw stream?” If the answer involves a support ticket, you do not own it.

“What does the API actually expose?” Live positions, historical traces, zone events, alert webhooks. A platform that gives you a dashboard but not events cannot be built on.

“How does this connect to my access control, HR system and CMMS?” The value compounds when location joins identity and work orders. Ask for a named customer doing exactly that.

On governance

“What are the default retention settings?” If raw traces are kept indefinitely by default, the vendor has not thought about the regulatory environment you operate in.

“Can I restrict who sees individual-level data versus aggregate?” Role-based separation between safety and management is the single most useful feature for getting workforce buy-in, and plenty of platforms do not have it.

“Can tracking be scoped to working hours and to site boundaries?” For a badge that leaves with the employee, this matters enormously.

On the relationship

“Who installs it, and who owns the survey?” Site surveys are where accuracy is won or lost. A vendor who subcontracts this and then blames the subcontractor is a known pattern.

“What is your support model in my time zone?” For sites in the GCC and India, a vendor whose support desk starts at nine o’clock in a European capital is a real operational risk.

The scoring exercise

Score each answer out of three. One for a vague answer, two for a specific answer, three for a specific answer with a reference customer you are allowed to phone. Any vendor scoring under twenty-four out of thirty-six should not be in your shortlist, no matter how good the demonstration was.

The pattern worth watching for is not incompetence. It is a vendor who answers every technical question precisely and every commercial question loosely. That combination usually means the technology works and the contract is where the pain will be.