The price structure, decoded
Every production monitoring quote is some mix of four layers. Making vendors split them is half the negotiation:
| Layer | What it covers | Indicative SA range |
|---|---|---|
| Sensing | Counters, current clamps, photo-eyes, reject gates — or PLC tag access where machines have one | Scales with machines; the R40K–R120K per-line band's biggest variable |
| Edge & connectivity | Gateway hardware, wiring, network or 4G/LoRaWAN links, load-shedding buffering | Included in per-line band; remote sites add connectivity cost |
| Platform & software | Dashboards, downtime reasons, alerts, reports — subscription or licence | Modest monthly fee per site/line; SaaS-only vendors price per machine per month |
| Services | Install, configuration, training, support SLA | The quiet differentiator — ask what's included vs billable |
Two pricing models dominate. Capex + subscription: pay for hardware and install once, modest ongoing platform fee — total cost transparent upfront. Pure SaaS per machine: low entry price per machine per month, but instrument-yourself assumptions and per-seat/per-feature escalations that only surface in year two. Neither is wrong; unpriced assumptions are.
What moves the number up or down
- Machine count and signal availability — a line with usable PLC tags meters cheaply; a 1990s line needs retrofit sensing, which adds hardware but no PLC project.
- Stop-reason granularity — automatic timing is cheap; operator terminals for tagging reasons add cost and add most of the value. Budget them.
- Integration ambitions — dashboards for the floor are one price; feeding SAP is another conversation. Phase it (see MES vs OEE).
- Load-shedding resilience — buffering and UPS-backed edge hardware should be standard in South Africa, not an option line. Quotes without it are quoting for a different country.
- Who installs — your electricians under remote guidance vs vendor crews on site. On multi-line rollouts this swings totals meaningfully.
Hidden costs to catch before signing
- Per-user or per-dashboard fees that punish exactly the wide visibility you bought the system for;
- Data egress/API charges if you later want your own data in your own BI tools — ask now;
- Sensor recalibration and replacement cycles, priced as service calls;
- "Integration-ready" claims that turn out to mean a CSV export — pin down protocols (OPC-UA/Modbus/MQTT) in writing;
- Minimum terms on SaaS pricing that outlive the pilot line's payback test.
How to buy this well
Start with one line — ideally the bottleneck — at fixed, written scope. Judge the system after 30 days of data on one question: did it find losses worth more than its annual cost? Most honest deployments clear that bar inside the first month (the downtime research explains why: unmeasured plants systematically underestimate their losses). Then scale line by line on evidence. This is how we structure our own OEE deployments — and why we publish indicative prices instead of "contact sales".