Key takeaways
- Table-stakes (2026): OPC UA / MQTT support, ISO 22400-aligned KPIs, reason-coded downtime, multi-site rollup, mobile.
- Differentiators: closed-loop CMMS integration, computer-vision-based cycle counting, recipe-anchored Performance, automated root-cause Pareto.
- Fluff: "AI-powered" buzzwords without specific use cases, vendor-locked dashboards, generic "industry 4.0" framing.
- Most demos pass the surface check — the real test is what happens in week 2 when the data starts moving.
- The features that matter most are reliability of data capture (resilient to network drops, PLC reboots) and ease of operator interaction.
Short answer: A useful OEE software features checklist separates table-stakes (must have), differentiators (worth paying more for), and fluff (marketing language that doesn't change the operational reality). In 2026, table-stakes include OPC UA/MQTT support, ISO 22400 alignment, reason-coded downtime, mobile access, and multi-site rollup. Differentiators include closed-loop CMMS integration and computer-vision cycle counting. Fluff is everything else dressed up to sound new. See also OEE vs Utilization.
Table-stakes features (do not buy without)
- OPC UA client support. Native, not via gateway. Must support Basic256Sha256 security.
- MQTT/Sparkplug B support. Especially for multi-site or unified-namespace strategies.
- ISO 22400-aligned KPI definitions. Auditable formulas, transparent inputs.
- Reason-coded downtime. Operators can tag stop events; reason codes feed Pareto analysis.
- Mobile operator interface. Phone or tablet for line operators to view current OEE and enter reason codes.
- Multi-site rollup. Compare lines across sites with normalized definitions.
- Reliable data capture. Resilient to network drops, PLC restarts, broker outages. The hidden killer of OEE platforms.
- Configurable for batch and discrete. Different time-state models, different Performance formulas.
Differentiators (worth paying more for)
- Closed-loop CMMS integration. Downtime event in OEE auto-generates a work order in CMMS, tracks the response, closes the loop. Most platforms claim this; few do it cleanly.
- Computer-vision cycle counting. Cameras as the cycle-count source where PLCs do not expose count data. Catches Performance loss invisible to PLC-only setups.
- Recipe-anchored Performance. Different ideal cycle time per SKU or recipe, not a generic line average. Essential for batch and high-mix discrete.
- Automated root-cause Pareto. Surfaces the dominant loss per line per shift without manual analysis.
- Operator-facing line view. A real-time, simple view that operators actually look at, not just management dashboards.
- Open API. REST/GraphQL with documented endpoints for integration with ERP, MES, BI tools.
Marketing fluff to filter out
- "AI-powered" with no specific use case. AI for what? Anomaly detection? Forecasting? Vague AI claims are usually marketing dressing.
- "Industry 4.0 ready." Means nothing specific. Ask what protocols it supports and what the data flow looks like.
- "Real-time" without latency numbers. Real-time at what cadence? 1 second, 5 seconds, 1 minute? Latency matters; vague claims do not.
- Vendor-locked dashboards. Pretty UI that exports nothing meaningful. Demand SQL access or open API.
- "Plug and play" claims. No industrial integration is truly plug-and-play. Ask about the typical implementation timeline.
What to test in a demo
- Show me a real customer's data, not a sandbox. Sandbox demos hide the messy edge cases.
- Disconnect the network for 5 minutes. Watch what happens to data capture and recovery.
- Add a new SKU with a different ideal cycle time. See whether the platform handles it without engineering intervention.
- Push a downtime reason code from the operator interface. See whether it appears in the Pareto within seconds.
- Ask for the SQL schema or API docs. Open access vs vendor lock-in is visible immediately.
Questions for procurement
- What is the typical time from contract to first production data?
- How is the platform priced (per line, per site, per signal)? Watch for hidden per-tag pricing.
- What is the SLA on data capture? On dashboard availability?
- What happens to my data if I leave the vendor?
- How is the platform updated? Cloud auto-update or on-prem with manual rollouts?
What the right OEE platform looks like in 2026
A platform with native OPC UA and MQTT support, ISO 22400-aligned KPIs, resilient data capture, closed-loop CMMS integration, mobile operator interface, and open API. Without those, you are buying a dashboard, not an OEE system.
Fabrico's OEE module covers the table-stakes and most of the differentiators: native OPC UA + MQTT, ISO 22400 alignment, closed-loop CMMS, recipe-anchored Performance, mobile operator view, REST API.
See how Fabrico captures this automatically — explore OEE for manufacturing or book a demo.
Related reading
Frequently asked questions
Is closed-loop CMMS integration really a differentiator?
Yes. Most platforms claim it; few do it cleanly. Test it in the demo by simulating a downtime event and verifying the WO appears in the CMMS with the right asset, time, and reason code.
Do I need computer vision for cycle counting?
Only if PLCs do not expose part count or reliable signals. Cameras catch cycles invisible to PLC-only setups, especially in older equipment.
Is mobile access really table-stakes?
Yes. Operators carry phones or use tablets. A desktop-only OEE platform is unusable on the line in 2026.
How much should I pay?
SMB plant with 5-10 lines: typically €40,000-€120,000 in year one (license + implementation + internal labor). Multi-site enterprise: much higher.
What is the biggest hidden risk?
Data capture reliability. Pretty dashboards hide ugly data quality. Test resilience to network drops and PLC restarts before signing.
Zaloguj
Blog
