RTLS pilot success criteria — what proves the business case.

Proof of Concept (PoC) vs Proof of Value (PoV) — which one do you need.

Proof of Concept answers "can this technology physically work in our RF environment?" with a controlled lab or single-bay test against defined success criteria, 2–4 weeks on rental kit.

Proof of Value answers "does it move the business KPI?" with a fully operational pilot in one area, line or ward, measured against a baseline, 8–16 weeks.

Skipping PoV and going direct to rollout is the most common cause of stalled deployments.

The five non-negotiable pilot scoping rules.

1. Baseline measurement before kit lands. If you cannot say what current WIP accuracy / asset search time / OEE is, you cannot prove the pilot moved it.

2. Production load, not demo load. Single-line, full shift, real operators.

3. Acceptance criteria signed before contracts. Written go/no-go thresholds for accuracy, read rate, latency, integration and adoption.

4. Integration proof to at least one core system. WMS, MES, EMR or ERP — pick one and prove the event flow end-to-end.

5. Named operations sponsor with decision authority. Without an empowered operations lead the technology will work and adoption will not.

Typical pilot timeline and cost ranges.

PoC: 2–4 weeks, €15,000–€40,000. Output: go / no-go on the technology.

PoV: 8–16 weeks, €40,000–€120,000. Output: board-ready business case.

Pilot: 12–20 weeks, €80,000–€250,000. Output: scoped rollout plan and TCO model.

Production-grade pilot gates: baseline, load, integration, adoption, decision

Vendor demos are designed to win a deal. Production-grade locating pilots are designed to expose edge cases before spend scales. Best practice is to separate Proof of Concept (can the radio work in our RF environment?) from Proof of Value (does locating move a business KPI under real operations?), and to refuse rollout until both have written gates.

Non-negotiable scoping rules. Baseline the KPI before kit lands. Run on a full shift with real operators and peak traffic. Sign accuracy, read-rate, latency, uptime, integration and adoption thresholds before contracts. Prove an event into at least one core system (WMS, MES, EMR or ERP). Name an operations sponsor with authority to change SOPs. Time-box the work (PoC 2–4 weeks; PoV typically 8–16 weeks).

Industry KPI examples. Hospitals: equipment search time, rental leakage, room-level assignment accuracy, duress response location quality. Manufacturing: WIP age accuracy, station presence, tool find time, safety-zone false-negative rate. Logistics: portal read rate, dock-to-stock, inventory accuracy, yard dwell. Pick one to three KPIs — not fifteen.

What “good enough accuracy” means in a pilot. Define percentile and conditions: e.g. “95% of equipment finds within the correct room during weekday peak” or “95th percentile error ≤30 cm in Cell 3 with line running.” Lab median numbers without load are not acceptance criteria.

Exit decisions. Go = gates met and TCO still holds. Repair = radio OK but density/integration/SOP wrong. No-go = wrong technology or no operational owner. Purchase hardware only at the pilot-to-rollout gate; keep rental through PoC/PoV where possible so a no-go does not leave you owning the wrong kit.

Evidence pack the steering group should see at the pilot gate

At go/no-go, present: baseline vs pilot KPI charts; accuracy/read-rate distributions (not single averages); integration event success counts; adoption metrics (active users, SOP compliance); issues log with severity; updated five-year TCO; and a rollout recommendation with explicit risks. Anything less invites a politics-led decision.

Include a short “what we still do not know” section — weekend patterns not seen, seasonal SKUs, future layout changes. Honest uncertainty builds more credibility than a perfect green slide. If gates were missed narrowly, propose a time-boxed repair plan rather than an automatic estate rollout.

Sample go/no-go thresholds you can adapt

Examples (adapt, do not copy blindly): room-level assignment ≥95% correct on peak weekday samples; portal read-rate ≥97% on designed pallet flows; 95th percentile UWB error ≤30 cm in the pilot cell with line running; event delivery to MES/WMS ≥99% within agreed latency; ≥80% of target operators completing the new SOP weekly by week 6. Write your numbers with operations before the vendor arrives on site.

Pair each threshold with a measurement method and a sampling plan. Thresholds without methods become arguments; methods turn them into engineering.

Rental kit and kill-switches protect capital

Keep hardware on rental or supplier-financed terms through PoC and PoV wherever possible. Pair that with explicit kill-switches at each gate so a no-go does not leave the organisation owning the wrong anchors. Capitalise only when go/no-go evidence and the five-year TCO still hold under production-load numbers.

Field note

Publish the measurement methods beside every threshold so go/no-go stays an engineering decision, not a debate.

Archive the evidence pack with the steering minutes for the next site rollout.

Keep the evidence pack auditable for the next site.