A university estates team recently shared a tender for occupancy monitoring across 14 buildings. One line read: "The system must provide accurate real-time occupancy data." Every one of the seven bidders confirmed compliance. Six of them could not have passed a basic acceptance test at the main library entrance, where a revolving door, a glass atrium and afternoon sun create exactly the conditions where cheap counting hardware falls apart. The tender never asked how accuracy would be verified, so nobody had to prove it.
This is the pattern in most public and institutional procurement of real time occupancy monitoring: the requirement is written as an outcome ("accurate data") rather than as a testable obligation. The result is that price decides the award, and the buyer discovers the difference between 88% and 98% accuracy only after go-live, when the safety officer stops trusting the live occupancy figure and goes back to manual counts.
Almost every vendor brochure claims "up to 99% accuracy." That phrase is unenforceable. Your tender should instead require a contractually guaranteed minimum counting accuracy, verified per counting point after installation. As a reference: serious providers will commit to a minimum of 96% in the contract, with 98–99% typically achieved in practice where lighting, entrance layout and visitor behaviour allow. If a bidder refuses to put a floor in writing, that tells you what their field performance looks like.
Specify the verification method too. A workable clause: manual validation counts of at least 200 passages per entrance during representative traffic, compared against system counts, before final acceptance and payment milestone release. Here is what implementers know and procurement teams rarely ask about: accuracy is not one number per building. It is per entrance, and it degrades in predictable places — wide entrances where groups walk abreast, doors near escalators, entrances where people linger with trolleys or prams. Require the bidder to identify high-risk entrances during a mandatory site survey and state expected accuracy per location, not as a site average that hides the problem doors.
The single most expensive mistake in occupancy tenders is buying a closed system where the analytics platform only works with one manufacturer's hardware. Sensors have a lifecycle of five to ten years; your data platform, dashboards and integrations should outlive several hardware generations. If the two are welded together, every future hardware refresh becomes a full re-procurement.
Write a hard requirement for a device-independent, sensor-agnostic platform: the software must ingest data from multiple sensor manufacturers and technologies (stereo vision, time-of-flight, thermal, Wi-Fi-based where appropriate). Vemco Group's platform is built this way deliberately — the software layer works across sensor brands — which also matters mid-contract. Universities and municipalities frequently inherit buildings with existing counters from earlier projects. A sensor-agnostic requirement lets you keep functioning hardware and consolidate everything into one live occupancy view, rather than ripping out working devices to satisfy a vendor lock-in.
Occupancy data earns its budget when it flows into other systems: BMS for demand-controlled ventilation, room-booking platforms, digital signage showing live capacity, BI tools for the estates department. Require a documented REST API with real-time or near-real-time endpoints, webhook support for threshold alerts (for example, occupancy exceeding 85% of permitted capacity), and named integration references. Vemco's VemFusion layer, for instance, exists precisely to push count data into BI, ERP and other enterprise systems — ask any bidder to demonstrate an equivalent live integration during evaluation, not just describe one.
Also score API rate limits and latency. "Real time" in some products means a five-minute polling cycle. For safety-capacity use cases in lecture halls or event venues, specify maximum acceptable latency in seconds and make the bidder commit to it.
A pragmatic scoring model for occupancy tenders weights verified capability over price. A pattern that works: 30% price, 25% contractually guaranteed accuracy and acceptance-test terms, 20% integration and data terms, 15% references from comparable environments (ask specifically for multi-building or campus deployments, not retail-only references if you are a university), and 10% support model and SLA. Require at least two reference customers you may contact directly, and ask them one question above all: what happened when a sensor started drifting — did the provider detect it, or did you?
That last question matters because sensor health monitoring is where providers genuinely differ. A platform processing count data at scale — Vemco handles tens of millions of counts daily across its customer base — has automated anomaly detection that flags a miscounting entrance within hours. A small integrator finds out when you complain. Put proactive data-quality monitoring and a response-time SLA for sensor faults into the requirements explicitly.
For multi-building estates, structure the award as a pilot phase (two or three buildings, including your hardest entrance) with contractually fixed unit pricing for the full rollout — and an exit right if the pilot fails acceptance testing. This keeps competitive pressure on the winner through delivery and gives you real accuracy data before committing the full budget. Bidders who are confident in their field performance accept this structure readily; the ones who push back are telling you something.
If you are drafting a tender or RFP for real time occupancy monitoring and want a second pair of eyes on the technical requirements — accuracy clauses, acceptance test design, integration specifications or hosting terms — the Vemco Group team reviews tender documents regularly and can share requirement language that has held up in real procurements. Get in touch at vemcogroup.com/contact-us before your tender goes out, not after the bids come back.