A well-designed productivity dashboard turns raw data streams into concrete shop-floor decisions. This guide shows how to build a productivity dashboard in your OEE software, from auditing your data sources through piloting and scaling, so you can improve OEE without hiring. You'll learn which KPIs to track, how to connect CNC and ERP/MES data, and which visualizations to prioritize for each role (operator, supervisor, management).
TL;DR:
Prioritize three KPIs: availability, performance, quality — measuring OEE correctly reduces unplanned downtime by 10–30% depending on the shop.
Automating ingestion from CNC (G-code, OPC-UA) and syncing with ERP keeps part-count discrepancies under 1% and cuts manual entry.
Run a pilot on 2 representative machines, apply time-based alert rules (30-minute threshold), then scale over 6–8 weeks with fast iterations.
Start by cataloging every source that touches production: CNC controllers (Fanuc, Siemens Sinumerik, Heidenhain), PLCs, MES history, ERP orders (job/work order number), and operator entries. Note the output format for each source: CNC logs (NC files or streaming), CSV files, SQL feeds, REST APIs, or industrial protocols (OPC-UA, MQTT). For small shops, an edge gateway publishing JSON over MQTT is often enough.
Also include capture devices: spindle-state sensors, part-presence sensors, tool counters, and barcode readers. These give you the machine status (in cycle, paused, stopped) needed to calculate OEE.
Time synchronization is the key. Verify that every source provides timestamps in UTC or a consistent local time, with at least one-second precision. If your CNCs produce logs without usable timestamps, you'll need an edge device to timestamp at ingestion. An initial audit helps identify drift: clock offsets, skipped logs, or mismatched formats.
For a detailed checklist on auditing your data before deployment, see the OEE data readiness audit.
Network infrastructure: dedicated VLAN for machines, QoS to prioritize telemetry, firewall configured for OPC-UA/API.
Time-series storage: a time-series database (InfluxDB, TimescaleDB) or direct ingestion into your OEE software.
Skills: a production engineer, an IT/OT contact (or vendor), and an operations champion to validate business rules.
Governance: a data owner, a validation procedure, and alert playbooks.
Per the OECD's productivity manual, clearly defining units of measurement and scope is the first step toward comparable indicators; read the guide to frame your definitions: Measuring Productivity - OECD Manual.
Automatically collect your OEE data — availability, performance, quality — directly from your CNC controllers, with no manual entry and no signal wiring.
Discover JITbase Machine MonitoringOEE (Overall Equipment Effectiveness) is classically calculated as: Availability × Performance × Quality. Define each term operationally:
Availability = (planned time − unplanned downtime) / planned time
Performance = theoretical production time / actual machine cycle time
Quality = conforming parts / parts produced
Specify how planned stops are treated (scheduled maintenance, breaks): these times should be excluded from the denominator if the goal is to assess operational performance during planned periods.
Track MTBF and MTTR for reliability analysis. To derive cycle time from a CNC program, use G-code extraction and tool-path simulation (feed rate, G1/G2 moves). For a deeper look at the technical method for estimating cycle time from code, see the article on extracting cycle time from G-code.
Example calculation rules:
Partially conforming part: count as non-conforming in Quality, but track defects by code for analysis.
Stop < 30 s: filter as noise if your sensor has instability issues.
Academic research on dashboard design recommends using a continuous time filter to compare performance by time slot; see best practices for dashboard design (UCSF): Tableau dashboard best practices.
Operator: machine status, remaining parts in the run, estimated cycle time, next action.
Shop supervisor: OEE by machine, bottlenecks, workload by station.
Management: OEE summary by line, weekly/monthly trends, downtime ROI.
Combine CNC shop-floor KPIs with operator efficiency KPIs to avoid attributing every loss to the machine alone.
Create separate pages: one for the operator (tablet or station screen), one for the supervisor with drill-down, and one summary for management. Each view should answer a clear operational question: what should the operator do right now? Where are today's bottlenecks? Separating views reduces "noise" and improves decision-making.
For layout templates and UX examples, see the article on dashboards by role and the impact of connected workers.
A widget should enable an action: for example, clicking an alert launches a playbook (restart a machine, change tooling, contact maintenance). Favor drills: line OEE → list of stops → probable cause → recommended action.
Operator view: status indicator (green/yellow/red), average cycle time, bar showing the next 5 operations, maintenance call button.
Supervisor view: heatmap of stops by hour, top 5 causes of unavailability, workload by station (planned vs actual hours).
Management view: 7/30/90-day OEE trend, MTBF/MTTR, histogram of non-conforming parts.
To balance human workload, add a widget for balancing operator workload and connect it to a labor management system to adjust assignments in real time.
Three common options:
Direct connection to CNC/PLC via OPC-UA or native protocols (requires network access and expertise).
An edge gateway that collects and normalizes signals, then publishes to your OEE system via MQTT/REST.
Connectors/APIs to your ERP to sync orders, bills of material, and statuses.
For small shops without a full MES, lightweight alternatives exist for planning production; see our article on free and low-cost scheduling tools.
Map at minimum:
Order: job/work order number, part reference, planned quantity
Machine: machine code, status, serial number
Time: status timestamp, duration in seconds
Quality: part ID, conforming/non-conforming status, defect code
Human resource: operator, shift
Sync the job number between ERP and OEE software to avoid traceability gaps.
Set up validations: a daily consistency check, alerts on produced vs. counted parts. A reconciliation process should fix gaps, not hide them. For examples of tracking automation and KPI synchronization, see the article on automating production tracking.
For methodological guidance on dashboard design and governance, look to general best practices from the dashboard design literature referenced above (UCSF) — the same continuous-filter and consistent-definition principles apply to data mapping and integration contracts.
For complementary analytics capabilities suited to CNC shops, the reading on manufacturing analytics software helps you choose tools that will enrich your productivity dashboard.
Sync your machine data with your ERP/MES and track your work-order progress in real time, with no manual re-entry.
See JITbase Production MonitoringHeatmap: shows the critical downtime hours of the day.
Stacked bars: breakdown of stops (maintenance, adjustment, tooling).
Sparklines: OEE trend over the last 24 hours.
Status indicators: large at-a-glance KPIs (OEE, parts/hour, non-conforming parts).
A good widget answers a single question and invites an action.
Set up alerts with delays and escalation:
Primary alert: machine OEE < 70% for 30 minutes → operator message + 5-action checklist.
Escalation: if unresolved after 15 minutes → notify the shop supervisor.
Logs: keep the action chain for post-mortem analysis.
Example playbook for OEE < 70% for 30 min:
Check CNC messages and fault code.
Inspect tooling (5 min).
If it's a mechanical stop, call maintenance and display the expected intervention time.
Enter a repair order in ERP/maintenance.
To turn alerts into concrete actions and improve OEE, see our article on improving OEE.
Choose 2 representative machines (one critical, one average). Measure OEE and key metrics for 2 weeks before setting a baseline, then roll out visualizations and alerts for 6–8 weeks. Measure the performance gap, intervention time, and conformance rate. Iterate every 2 weeks.
For concrete real-time dashboard examples, see the real-time KPI dashboard guide.
Here are six frequent mistakes and how to fix them:
Bad timestamps: sync clocks (NTP) and timestamp at ingestion.
Poorly defined KPIs: document definitions (e.g., how planned stops are treated).
Too many non-actionable alerts: introduce a minimum delay and prioritization.
No role-based views: create separate pages for operators and management.
Reliance on manual entry: automate counting from CNC/counters.
Unsynchronized integrations: establish a daily reconciliation process.
Before adding more widgets, verify that every alert leads to a playbook. An alert with no action discourages teams.
To go further on designing interactive dashboards and configuring your widgets, consult the official documentation for your visualization tool (Power BI, Tableau) alongside the real-time KPI dashboard articles linked above.
Define a pilot scope, milestones, and success KPIs:
Scope: 2 machines + 1 line, 8-week pilot.
Success KPIs: relative OEE improvement (target e.g. +8–15%), 20% MTTR reduction, 90% reduction in manual entries.
ROI measurement: a calculator based on machine hourly rate, labor cost, rejected parts. For methods and models, see our guide on building an OEE software business case.
Train operators on the new screens and playbooks. Prepare a quick kit: a one-page cheat sheet per role, two 30-minute sessions, video support. Track adoption with metrics: number of actions launched from the dashboard, average response time.
After a successful pilot, plan a rollout in waves of 4–6 machines. Keep a 2-week feedback window after each wave to fix ERP/MES mappings or alert thresholds. To compare solutions and technical selection criteria, see the article on the best OEE software.
Quantify the return on investment of your productivity dashboard before rolling it out shop-wide.
Calculate My ROIA well-designed productivity dashboard in your OEE software helps you spot production losses and take fast action that improves OEE. Start with a data audit, define operational KPIs, pilot on a few machines, and iterate while measuring ROI before scaling up.
First check timestamp consistency (NTP sync) and the match between counted parts and produced parts across several hourly samples. Run an audit comparing CNC/PLC logs against ERP orders for three consecutive days: if the gap exceeds 1%, identify sources of loss (missing logs, order-mapping errors, faulty sensors).
Use an end-to-end test: start a known run, compare automatic counting versus operator entry, and fix data transformations before going live.
Reduce noise by imposing a minimum delay before notification (e.g., 30 minutes for OEE), prioritizing alerts by criticality, and fine-tuning thresholds per machine. Convert low-priority alerts into daily reports rather than immediate notifications.
Finally, pair every alert with a clear playbook: if an alert has no associated action, remove it or fold it into an analysis view.
There are two approaches: direct G-code extraction to estimate motion time (parsing G1/G2 commands and computing feed-based distances), or measuring via sensors/timestamps in production. G-code extraction gives a useful theoretical estimate for planning. For a detailed method and implementation example, see the article on G-code cycle time.
Adopt an adaptive granularity: minute/5-minute for the operator and local alerts, hourly slices for the shop supervisor, and daily/weekly for management. This approach enables fast action without flooding senior decision-makers with unnecessary detail.
In practice, start with 1-minute ingestion of machine states, then display aggregations at 5 minutes, 1 hour, and 1 day depending on the view and role.