Manufacturing data quality audit

Why run this

What this checklist is for

Operations reports efficiency, finance reports margin, and the two rarely agree. The disagreement is almost always a data quality problem rather than a disagreement about facts.

Every check below is about whether a number derived from the plant would survive being questioned by somebody who works there.

Run it with

Operations director and finance director jointly

And with

Data engineers

And with

A production manager who knows the floor

The checklist

20 checks, in the order we would run them

Tick only what you can genuinely evidence today. An item you intend to do is not an item you have done, and scoring yourself generously here only produces a comfortable number and an uncomfortable project.

Section 1

Capture

Whether the data reflects what happened.

Machine events carry start and end times, not only durationsDurations without timestamps cannot be assigned to shifts or reconciled to orders.
Machine clocks are synchronised to a reference and drift is correctedDevices drift, sometimes substantially. Events land in the wrong shift and the analysis is quietly wrong.
Downtime reasons are captured against a controlled taxonomyFree-text reasons produce a report nobody can aggregate and nobody acts on.
Scrap and rework are recorded at the operation, not only at the orderOrder-level scrap tells you it happened. Operation-level tells you where.
Manual entries are distinguishable from automated onesThey have different reliability and should be analysable separately.

Section 2

Completeness

Whether anything is silently missing.

Every production line and machine in scope is actually reportingCompare the reporting device list against the physical asset register.
Gaps in the event stream are detected and alerted, not silently toleratedA machine that stopped reporting looks exactly like a machine that stopped running.
Reconnection after an outage does not duplicate or lose eventsTest it deliberately, including out-of-order delivery. This is the failure most designs carry.
Buffer depth at the edge is monitored with a threshold alertSilent accumulation is the dangerous failure because everything looks fine.
Unclassified events are surfaced rather than excludedNew event types silently dropped from analysis is how coverage degrades.

Section 3

Reconciliation

Whether plant data agrees with the business systems.

Machine-reported production reconciles to ERP posted quantity within toleranceA persistent gap is a finding — usually unreported scrap or a counting method difference.
The tolerance is agreed and variances above it are investigatedAn unagreed tolerance means every variance is either ignored or argued about.
Equipment-to-work-centre mapping is maintained data rather than hard-codedPlant layouts change. A mapping in code changes only when somebody remembers.
Shift boundaries are handled correctly, including cross-midnightDeriving shift from a timestamp produces subtle errors that survive months of reporting.
Planned versus unplanned downtime is classified consistentlyIt is the denominator in availability and the most disputed definition in OEE.

Section 4

Use

Whether the data is actually informing decisions.

OEE definitions are agreed between operations and finance and publishedThe first question anybody asks is how it is calculated. A visible answer stops the number being disputed.
Plant performance and product margin appear in the same viewThis is what changes the improvement conversation from an argument to a business case.
A production manager can drill from a metric to the underlying eventsThree clicks or fewer. Anything more and the number gets disbelieved rather than investigated.
At least one operational decision in the last quarter used this dataIf nobody can name one, you have built reporting rather than capability.
Data quality itself is measured and reportedCompleteness, reconciliation variance and unclassified rate as standing metrics.

What your score means

Read this against the number above

These bands are deliberately blunt. The middle band is where most organizations honestly sit, and it is a perfectly reasonable place to proceed from — provided the gaps are written down with owners rather than carried as optimism.

0–59%Significant gaps

Do not proceed yet. More than four in ten items are unaddressed, and the ones that fail here are usually the foundational ones that make everything after them harder.

60–84%Mostly ready, with known gaps

Proceed on a defined scope, with the outstanding items written into the plan as risks with owners and dates. This is the most common honest position.

85–100%Ready

The remaining gaps are small enough to handle during delivery rather than before it. Confirm the unticked items are genuinely minor rather than simply unexamined.

Your score highlights automatically as you tick items above. Nothing is saved, sent or tracked — refreshing the page clears it.

Closing the gaps

If you could not tick these, start here

The four items below are the ones whose absence causes the most trouble downstream. If your unticked items include any of these, they are worth addressing before the rest.

Clock drift uncorrected

Fix this first. Events landing in the wrong shift undermine every downstream analysis and the error is invisible.

Reconnect untested

Test it with a realistic outage and out-of-order delivery. Silent double-counting is the failure that surfaces months later in a disputed report.

Free-text downtime reasons

Move to a controlled three-level taxonomy. It is the change that makes downtime analysis actionable rather than descriptive.

No reconciliation to ERP

Build it as a visible measure. A persistent gap is a finding worth having, not an inconvenience to reconcile away.

Want a second opinion on your score?

We will assess one line's data end to end and reconcile machine-reported production to your ERP for a closed period, which usually establishes the real gap quickly.

Talk through your result Read the related guides We reply to every message within one business day.

Keep going

Related checklists

business-central

manufacturingReadiness

Inventory count readiness checklist

Twenty checks before a physical count, so the adjustment tells you something rather than merely reconciling a number.

May 16, 2026 · 20 checksOpen

azure

manufacturingReadiness

Disaster recovery readiness checklist

Twenty checks on whether a manufacturer could resume production after losing its systems, measured in shifts rather than days.

March 21, 2026 · 20 checksOpen

What this checklist is for

Run it with

And with

And with

20 checks, in the order we would run them

Capture

Completeness

Reconciliation

Use

Read this against the number above

If you could not tick these, start here

Clock drift uncorrected

Reconnect untested

Free-text downtime reasons

No reconciliation to ERP

Want a second opinion on your score?

Related checklists

Inventory count readiness checklist

Disaster recovery readiness checklist

Tell us what you're trying to fix

Let’s get started on something great

Phone

Email