Every vendor demo shows a dashboard with an OEE gauge. That is the easy part. The hard parts are upstream and downstream of the dashboard: getting a true stop and count signal from a 1998 press and a 2024 packaging line with the same product, getting operators to classify stops in the 15 seconds before a restart, and getting production and maintenance to act on the Pareto every week. A system that fails any of the three produces a beautiful screen and no OEE improvement. This guide is organised around those three failure points, and the scorecard scores vendors on them.
What you get
- The buyer’s guide (PDF, 5 pages): the five categories of solution and where each fails (manual, historian, MES module, dedicated monitoring, IIoT platform), build vs buy decided honestly, the 10 requirement areas and their default weights, the four data-capture methods and their trade-offs, pricing models and the 3-year TCO, a seven-step evaluation process, nine red flags, and what a working deployment looks like.
- The RFP scorecard (Excel): 51 requirements in 10 categories, each with a must / should / nice priority, a “how to verify” method and a 0 to 5 score for up to 3 vendors; a Weights sheet; a Summary with category scores, overall weighted score, must-have failures, exclusion flag, rank and chart; a 3-year TCO sheet per vendor with cost per machine; a Pilot plan sheet with 7 success criteria to paste into the vendor proposal; and a Reference-call script.
The 10 requirement areas and what the must-haves protect
| Area | Default weight | Must-haves |
|---|---|---|
| Data capture and connectivity | 20% | Stop and count without PLC changes, 1-second resolution, no data loss on network drops |
| KPIs and analytics | 15% | Your time model, not the vendor’s; drill-down from Pareto to events |
| Operator experience | 15% | Reason entry under 15 seconds at the machine |
| Downtime and reason codes | 10% | 3-level configurable tree with planned / unplanned and six-big-loss attributes |
| Integration and API | 10% | Open, documented API |
| Deployment, security, IT | 10% | Outbound-only network design; you own the data |
| Alerts and workflows | 5% | |
| Vendor and services | 5% | Contactable references; a pilot with written success criteria |
| Pricing and TCO | 5% | 3-year total in writing |
| Scalability and multi-site | 5% |
Fifteen requirements are marked “must”. A vendor scoring 0 or 1 on any of them is excluded whatever the weighted total. This prevents a polished analytics layer from compensating for a capture method that will not work on a third of your machines.
The evaluation process that works
- Machine inventory and capture constraints (1 week): age, controller, PLC access, changeovers.
- Requirements and weights (1 week): scorecard adapted, must-haves confirmed with production, maintenance and IT.
- Long list to short list on documents (2 weeks): API documentation and security architecture read before any demo.
- Scripted demo on your data (1 week): your time model, your reason tree, a network unplug, a Pareto drill-down.
- Pilot on 2 to 3 machines (4 to 6 weeks): scored against the 7 success criteria.
- Reference calls and TCO (1 week): two references per finalist, TCO sheet completed.
- Decision and rollout plan (1 week): machines per month, one tree, one KPI definition.
Who this is for
Plant and operations directors starting an OEE or machine-monitoring selection, CI and industrial engineering leads writing the requirements, and IT / OT teams who need the security and integration questions asked before the contract. It complements our OEE software cost guide and the downtime tracking software comparison.
Download the guide and the scorecard
Fill in the form and both files are sent to your e-mail address.
Guided pilot
A note on neutrality
The scorecard is written to be fair to all five solution categories: an MES module or an IIoT platform that meets the must-haves on capture and operator experience scores as well as anything else. TEEPTRAK is a dedicated production monitoring system with sensor-based capture that works without PLC changes, and we are comfortable being scored on the same 51 lines. Across the 450+ plants we have equipped, the pattern of a successful deployment is consistent: capture validated on every machine type in the first two weeks, operators classifying stops by week three, the first Pareto review before the end of the pilot. One documented case, Hutchinson, moved OEE from 47% to 72% once losses were visible and worked in sequence. See how TEEPTRAK works.
Related reading: the OEE Buyer’s Kit 2026 (benchmarks and pilot plan for OEE software specifically) and the OEE vendor evaluation scorecard article, which this pack turns into a working Excel with weights, a must-have gate and a 3-year TCO.
Related reading: the OEE Buyer’s Kit 2026 (benchmarks and pilot plan for OEE software specifically) and the OEE vendor evaluation scorecard article, which this pack turns into a working Excel with weights, a must-have gate and a 3-year TCO.
Frequently asked questions
What should be in an RFP for production monitoring or OEE software?
Ten requirement areas: data capture and connectivity, KPIs and analytics, operator experience, downtime and reason codes, alerts and workflows, integration and API, deployment and security, vendor and services, pricing and TCO, scalability and multi-site. The scorecard ships 51 requirements across them, each with a priority, a verification method and a 0 to 5 score per vendor.
How is the weighted score calculated?
Each requirement has a priority weight (must 3, should 2, nice 1) and a vendor score from 0 to 5. The category score is the weighted sum divided by the weighted maximum; the overall score applies your category weights (default: capture 20%, KPIs 15%, operator experience 15%, reason codes 10%, integration 10%, security 10%, the rest 5% each). Any must-have scored 0 or 1 excludes the vendor whatever the total.
How long should a pilot be and what are the success criteria?
Four to six weeks on 2 to 3 machines, including one legacy machine without PLC access. Seven criteria: stop detection above 98% versus observation, count within 2% of manual, more than 90% of stops classified by operators within the shift, OEE by hand within 1 point of the system, a Pareto review held in week 3 with two actions launched, reason entry under 15 seconds, and the IT and OT security review closed.
Should we build our own OEE dashboard instead of buying?
Build when all machines expose data, you have an in-house software team and the scope is dashboards, not operator interaction. Buy when the fleet is mixed, operators must classify stops at the machine and several plants need one definition. The common answer is hybrid: buy the capture and operator layer, keep BI and ERP integration in-house through an open API. The guide has a section on estimating the build option honestly.
0 Comments