Power BI for Manufacturers, built around OEE.
Petronella Technology Group, Inc. builds Microsoft Power BI dashboards for manufacturers that turn ERP, MES, PLC, and QMS data into one secure, real-time view of plant performance. OEE, downtime by cause, scrap, supplier on-time, cost-per-unit — on the same screen, refreshed every few minutes.
TL;DR · What Power BI dashboards do manufacturers need?
Six. A Plant Manager Daily opening with yesterday's OEE and top three downtime causes. An OEE by line and shift view for asset and crew comparisons. A Supplier Scorecard ranking vendors by on-time and quality. A Quality & CAPA dashboard tied to the QMS. A Production vs Demand view pairing MRP plan against actual output. And a Cost-per-Unit dashboard blending labor, material, and overhead into a single contribution-margin number per SKU. Every other manufacturing dashboard worth building is a slice of those six.
Most manufacturers are flying blind on plant performance.
The ERP knows what was sold. The MES knows what was run. The PLCs know what stopped. The QMS knows what failed inspection. The MRP knows what is supposed to be made next week. None of those systems talk to one another in a way the plant manager can use on a Monday morning. The shift report explains what already happened on Friday, and by the time anyone acts on it, the same downtime cause has already eaten Tuesday.
Power BI for manufacturing closes that gap. Petronella Technology Group, Inc. pulls each source — ERP, MES, PLC, QMS, MRP, time-and-attendance, supplier APIs — into a single Microsoft semantic model. We model OEE the way ISA-95 and ISO 22400 expect it to be modeled. We refresh it every few minutes for the shop floor and every 30 minutes for management. And we put the right view on the right screen for the plant floor, the front office, and the boardroom.
This page is for the COO, VP of Operations, or Plant Manager at a US manufacturer somewhere between $50M and $250M in revenue who is tired of spreadsheets and tired of shift reports that explain the past. If that's you, you don't need another BI vendor presentation. You need a dashboard that opens with last shift's OEE and the three things the next shift can fix.
Nine manufacturing KPIs every Power BI model should ship with.
These are the measures that move profit. We build each one as an explicit DAX measure with a tested formula, a documented data lineage, and a drill-down path to the underlying records.
Overall Equipment Effectiveness
OEE = Availability × Performance × QualityThe headline plant number. Each factor drills to the underlying scheduled time, actual run time, ideal cycle time, and reject counts. World-class is 85%; most plants live in the 55-70% band when measured honestly.
Downtime by Cause
Pareto by reason codeTotal downtime broken out by reason code, sorted big-to-small. Drill from the chart to the individual events with timestamps, operator notes, and the PLC stop code that triggered them.
Scrap and Rework Rate
Defects ÷ Units ProducedBroken out by line, shift, SKU, and supplier lot. Pairs with the Quality & CAPA dashboard to surface drift before it becomes a customer return or a CAR.
Throughput vs Takt
Units ÷ Available TimeReal output rate compared to the takt time the schedule was built on. The gap between the two is where Performance loss lives in the OEE calculation.
Supplier On-Time Delivery
On-Time Receipts ÷ PO LinesJoined against your ERP receiving data with a configurable on-time window. The Supplier Scorecard ranks every vendor on a single page so purchasing has an evidence-based conversation, not an opinion-based one.
Inventory Turns
COGS ÷ Avg InventoryCalculated by SKU class and location with slow-mover and dead-stock flags. Tied back to MRP so you can see how forecast accuracy is driving your working capital.
First-Pass Yield / Defect Rate
Pass ÷ Inspected, by stepCalculated per inspection point so you can see exactly where in the process value is being destroyed. Drills to the operator, the line, and the supplier lot.
Schedule Adherence
Planned vs Actual quantityPer work order, with on-time start, on-time complete, and the variance between MRP-planned and actually-made quantities. Surfaces the orders that quietly slip every week.
Labor Utilization
Direct ÷ Paid HoursDirect hours divided by paid hours, broken out by department and shift. Tied to time-and-attendance and labor reporting on the MES so the number isn't a manager's estimate.
Every system in your plant, connected through one model.
We bring the data in where the data lives. No "rip and replace." No requirement that you migrate ERPs first. Power BI sits on top of what you already run.
Real-time when it matters, scheduled when it doesn't.
Manufacturers ask for "real-time dashboards" and then discover the operational database can't take the query load. The right answer is a refresh mode per dashboard, not one mode for the whole tenant.
Import mode — the default for management views
Data is copied into the Power BI semantic model on a schedule. Refresh frequency is typically every 30 minutes for shift-level reporting, hourly for management, and once or twice a day for executive roll-ups. Import is fast, cache-friendly, and the only mode where every DAX calculation is fully available with no source-side limitations. For most manufacturing dashboards, this is the right default.
DirectQuery — for near real-time line views
The query runs against the source database every time a user interacts with the visual. There is no copy and no refresh schedule. The trade-off is that complex DAX measures translate into complex SQL, and the source database has to be tuned to handle the dashboard's query pattern. We use DirectQuery selectively — usually for a single live OEE tile rather than the whole report.
Direct Lake on Microsoft Fabric — the modern shop-floor pattern
For clients on Microsoft Fabric, Direct Lake reads Delta tables in OneLake directly — no import, no DirectQuery, no caching. It delivers import-class performance at DirectQuery-class freshness. For manufacturers landing PLC telemetry into a Fabric lakehouse, this is now the default shop-floor pattern. Microsoft requires Fabric capacity (F2 or higher, or Power BI Premium P1 or higher) for Direct Lake; Pro and PPU alone are not sufficient.
Hybrid models
You can mix all three in a single report. A typical Plant Manager dashboard has Import tables for ERP and finance data, DirectQuery against the MES for current-shift output, and Direct Lake against a Fabric lakehouse for PLC time series. We design the mix so the report opens fast and the right tiles update at the right cadence.
Six manufacturing Power BI dashboards we build over and over.
These are the production reports inside almost every engagement. Each one has a defined audience, a defined refresh cadence, and a documented set of actions a supervisor can take when the dashboard surfaces a problem.
Plant Manager Daily
The first thing the plant manager opens in the morning. Yesterday's OEE, the three biggest downtime causes, scrap rate vs trailing week, supplier exceptions, and any work orders that slipped overnight.
OEE by Line & Shift
OEE for every line and every shift on one page. Each cell drills into Availability / Performance / Quality components. Used in daily Gemba and weekly continuous-improvement reviews.
Supplier Scorecard
Every supplier ranked by on-time delivery and quality (PPM defective). Filterable by category, item class, and dollar spend. Drives quarterly business reviews and supplier corrective-action conversations.
Quality & CAPA
Open non-conformances, open CAPAs by age, repeat-offender SKUs, customer-complaint trend, and the audit findings still in remediation. Joined to the QMS so the dashboard and the system of record never disagree.
Production vs Demand
MRP plan compared to actual output, work-order completion variance, and finished-goods inventory against forecast demand. Surfaces the "we built it but we didn't need it yet" and "we needed it but we didn't build it" gaps that hide in the ERP.
Cost-per-Unit
Labor, material, overhead, and scrap cost blended into a single contribution-margin number per SKU. Snapshots month-over-month so you can see margin compression as it happens, not at month-end close.
Discovery · Model · Build · Secure · Train · Manage.
A repeatable engagement model. The Manufacturing OEE Pack is the smallest unit; Foundation and Enterprise extend the same approach with more modules and more plants.
Discovery and OEE Definition
A short, evidence-based discovery: which ERP, which MES, what data feeds from the floor, who owns each system, and exactly how this plant defines OEE today. We write the definition down so the dashboard and the management team agree on the math.
Data Model & Dashboards
We build the semantic model in Microsoft Fabric or Power BI Desktop, write the DAX measures, and ship the Plant Manager Daily, OEE by Line, and Supplier Scorecard. Each dashboard goes through one technical and one user-acceptance review before sign-off.
Security, Training & Run
Row-Level Security per plant, sensitivity labels for cost and margin data, audit-log forwarding to your SIEM, and operator and analyst training. Optional managed-reporting retainer so Petronella Technology Group, Inc. runs the platform and adds measures as the business changes.
Anomaly detection and natural-language Q&A — without sending data outside the boundary.
Microsoft has invested heavily in Copilot for Power BI, and for many use cases it works well. For manufacturers handling supplier IP, customer technical data, or DoD-controlled information, sending dashboard prompts to a public AI service is a non-starter. Petronella Technology Group, Inc. runs a private AI fleet (Penny) that delivers the same natural-language experience without the data ever leaving the boundary you control.
What that looks like in practice:
- Anomaly detection on shop-floor data. A nightly batch flags lines whose downtime is two standard deviations above their 30-day mean, scrap rates drifting outside SPC limits, and supplier delivery windows shifting beyond their historical band. The signal lands on the Plant Manager Daily and in a Telegram channel before the 6 AM shift huddle.
- Penny-on-your-data for plant-floor Q&A. Operators and supervisors ask "what's our OEE on Line 3 this shift?" or "show me every CAPA still open more than 60 days" in natural language. Penny answers from the Power BI semantic model, not from a public LLM, and every query is auditable.
- Demand forecast review. We compare forecast accuracy against actual demand by SKU class and surface the items where the forecast has degraded enough to matter to inventory and labor planning.
For the architectural background, see our AI services overview and our practical guidance on private versus public AI in regulated environments.
Five plant profiles we routinely build for.
Every plant is different, but the patterns repeat. Here's how the Manufacturing OEE Pack maps onto five common scopes.
100-employee custom job shop · one location, mixed-model production
Custom job shops live and die on schedule adherence and rework. The build starts with ERP order data (typically Global Shop Solutions, JobBoss, or Made2Manage), labor reporting from the shop-floor terminal, and CAD/router data for first-article QA. The dashboards focus on order-by-order margin, on-time completion, and rework root cause. Many job shops never had real OEE before because the product mix was too varied; we model "available time vs paid time" and quality at first inspection so the leadership team has a comparable number across job classes.
250-employee multi-line plant · mid-market with a real MES
Mid-market plants with Plex, Tulip, or Ignition have rich shop-floor data but rarely combine it with ERP and supplier data in one view. The OEE Pack lands first: Plant Manager Daily, OEE by Line, Supplier Scorecard. The Foundation engagement that follows usually adds Quality & CAPA (joined to the QMS) and a Cost-per-Unit dashboard. Refresh strategy is typically Direct Lake on Fabric for the floor tiles and Import for the management views.
Multi-site $250M industrial · consolidated reporting across 4-8 plants
Multi-site reporting only works if every plant agrees on the OEE math. The engagement starts with a definition workshop — one document signed by every plant manager — before any dashboard work. The semantic model has a single Plant dimension and Row-Level Security per site. The COO sees the roll-up; each plant manager sees their own data plus one comparable peer. Quarterly business reviews stop being a fight about whose number is "really" right.
Contract manufacturer doing CMMC reporting · DoD supplier with CUI
Contract manufacturers and machine shops on the Department of Defense supply chain need Power BI deployed inside a CMMC-aligned boundary because the production data can contain Controlled Unclassified Information (CUI). The two practical paths are Microsoft GCC High (full Microsoft stack inside a US-government tenant) or a Petronella encrypted enclave pattern that costs less and gives you tighter control. Either way, Petronella Technology Group, Inc. is CMMC RPO #1449 with four CMMC Registered Practitioners on staff, so the boundary and the dashboards come from the same team. Pair with our CMMC compliance practice and ComplianceArmor for documentation.
Single high-mix CNC shop · ten machines, owner-operator
Smaller shops often skip a formal MES and run on spreadsheets, paper traveler routings, and the ERP. A pragmatic Power BI build pulls ERP order and time-card data, an OPC UA bridge on the most critical CNC controllers, and quality data from whatever the inspector enters at the part. The dashboards stay tight: a one-page owner daily that shows revenue, throughput, and the two machines that are losing money. No heroics; just data the owner can act on before the next quote goes out.
How a Petronella Power BI engagement compares.
| Attribute | Petronella Technology Group, Inc. | MES-vendor built-in dashboards | Hire an internal data analyst | Migrate from Tableau / Looker |
|---|---|---|---|---|
| Time to first working OEE dashboard | ~4 weeks (OEE Pack) | Same day, but scoped to that MES only | 6-12 months including hiring | 3-6 months migration + retraining |
| Cross-system data (ERP + MES + PLC + QMS) | Yes, single semantic model | No, MES data only | Depends entirely on the hire | Yes, but custom ground-up |
| CMMC / ITAR boundary awareness | Yes — RPO #1449, four CMMC-RP | Generally no | Rarely | Generally no |
| Private AI for plant-floor Q&A | Penny — data stays on your boundary | Vendor-specific Copilot if any | Public LLM, data leaves boundary | Public LLM, data leaves boundary |
| Total first-year cost | Fixed-fee, 100% upfront, scoped per stack | Bundled in MES subscription | Senior BI analyst salary + benefits + tools, recurring | Migration cost + ongoing license |
| Ongoing maintenance | Optional managed retainer | Vendor manages | Internal staff | Internal staff |
| Microsoft stack alignment | Native — Power BI, Fabric, Azure | Varies by vendor | Whatever the analyst knows | Migrating off the alternative |
Manufacturing OEE Pack · Foundation · Enterprise.
Fixed-fee, payment due 100% upfront at contract execution. Pricing depends on the ERP, MES, PLC, and QMS stack on site and the number of plants in scope. Request a quote and we will scope against your environment.
Manufacturing OEE Pack
The starter scope. Discovery, OEE definition workshop, semantic model build, three production dashboards (Plant Manager Daily, OEE by Line, Supplier Scorecard), Row-Level Security per plant, sensitivity labels, audit-log forwarding, and operator/analyst training. Delivered in approximately four weeks for a single plant on a clean ERP and MES stack.
Foundation
OEE Pack plus three additional dashboards (Quality & CAPA, Production vs Demand, Cost-per-Unit) and integration with your QMS and time-and-attendance system. Six to eight weeks. Includes the Anomaly Detection module on the Petronella AI fleet.
Enterprise / Multi-Plant
Foundation scope rolled out across multiple plants on a phased schedule with a consolidated COO view, per-plant Row-Level Security, executive scorecards, and a managed-reporting retainer that covers Fabric capacity tuning, ongoing measure development, and 24x7 alerting on key KPIs.
Pricing notes
We deliberately do not publish hardcoded Power BI prices on this page. Engagement cost depends on how clean your data sources are, how many plants are in scope, whether you need a CMMC-aligned boundary, and whether the managed-reporting retainer is included. Request a quote and we will scope honestly against your environment. Payment terms are 100% upfront at contract execution — no splits, no net-15.
Power BI for Manufacturers — twelve common questions.
What Power BI dashboards do manufacturers need?
Can Power BI calculate Overall Equipment Effectiveness (OEE)?
Can Power BI pull data from an MES or a PLC?
How fresh is the data in a manufacturing Power BI dashboard?
Which ERP systems do you integrate with for manufacturing Power BI?
How long does a manufacturing Power BI build take?
How does Power BI handle multi-plant manufacturing reporting?
Can Power BI replace our MES vendor's built-in dashboards?
Is Power BI a fit for a CMMC or ITAR-regulated manufacturer?
Where does AI fit into a manufacturing Power BI program?
What does a Power BI for manufacturers engagement cost?
Do you train our staff to maintain the dashboards?
Related Petronella services
- Pillar: Power BI Consulting Raleigh NC
- Sibling: Copilot for Power BI Consulting
- Hub: Managed IT services for Raleigh businesses
- Hub: AI services and private LLM deployment
- Compliance: CMMC compliance — RPO #1449
- Compliance: ComplianceArmor® policy automation
- Cyber: Cybersecurity for manufacturers (IP protection)
- Team: The Petronella team
About the author
Craig Petronella is the founder and principal of Petronella Technology Group, Inc., a Raleigh, NC firm building Microsoft data, AI, and cybersecurity programs for US businesses since 2002. Craig is a CMMC Registered Practitioner (CMMC-RP), holds Cisco's CCNA and CWNP's CWNE certifications, and is a North Carolina-licensed Digital Forensic Examiner (License 604180-DFE). He is an Amazon #1 Best-Selling Author of 14+ cybersecurity books and an MIT Sloan alum in AI Implications for Business Strategy.
Petronella Technology Group, Inc. is CMMC RPO #1449 with four CMMC-RPs on staff (Craig Petronella, Blake Rea, Justin Summers, Jonathan Wood) and a 20+ year BBB A+ accreditation. Full bio at /about/craig-petronella/.
Ready to see your plant in real time?
Talk to Petronella Technology Group, Inc. and we will scope a Manufacturing OEE Pack against your actual ERP, MES, PLC, and QMS stack. Fixed fee, four-week delivery, no rip-and-replace.