Two teams can report different figures for what should be the same manufacturing KPI.

Production may show one result, finance another, and a local spreadsheet another. Each figure may be based on legitimate data, but the differences can make it difficult to know which one to use.

The difference usually comes from discrepancies in the underlying data, the definitions or calculation rules applied, or the point in time when the figure was produced.

These differences are common when manufacturing information moves across several plants, systems and teams. They do not necessarily mean one system is wrong; the same measure can produce different results when its data, rules or timing differ.

Adding another report or dashboard will not resolve that disagreement if nobody can explain how the figure was produced. The more useful place to start is understanding where the data begins to differ, what happens to it along the way and whether everyone is working from the same definition.

1. Why teams can report different figures for the same manufacturing KPI

A Key Performance Indicator, or KPI, is a quantifiable measure used to track performance against an important business objective. Not every operational metric needs to be a KPI; KPIs are the measures an organisation chooses because they matter most for understanding performance. Even when teams use the same KPI name, differences in what they include, when they record it, and how they calculate it can produce different results.

Take downtime as an example. Two plants may both report downtime as an important performance measure, but one may include certain planned stops while another excludes them. If they use different rules, their reported downtime figures are not directly comparable, even though both reports use the same label.

Production output can create similar problems. One source may record a unit as soon as it leaves the line, while another only includes it when the production order is confirmed. Rework, rejected products, or partially completed orders may also be treated differently.

None of this automatically means the underlying data is poor. The figures may have been produced using different data or rules.

The problem appears when those differences are not visible, and people assume that two figures labelled with the same KPI should match. Then time is spent checking the numbers rather than discussing what the performance result means.

So when two reported figures for the same KPI disagree, the first question should not be “Which system is wrong?” It should be “Are these figures based on the same data, definition, calculation and time period?”

2. Where ERP, MES, SCADA and spreadsheet data start to differ

Manufacturing information usually sits across systems that were designed for different purposes, so it is normal for each one to hold a different part of the operational picture.

An Enterprise Resource Planning (ERP) system may contain information about production orders, materials, inventory, planning and finance. A Manufacturing Execution System (MES) is closer to manufacturing operations and may hold more detailed information about what happened during production. Supervisory Control and Data Acquisition (SCADA) systems can monitor process conditions and equipment states, while local spreadsheets may contain targets, classifications, or adjustments not held elsewhere.

Those differences matter when you combine information from several sources to calculate or report a measure.

For example, a machine may record a stop immediately, while the reason is entered later in the MES. The ERP may receive updated information when the production order is confirmed, and a spreadsheet may then add or adjust something before the weekly report is prepared.

By the time management sees a KPI, its underlying data may have passed through several systems, calculations and manual steps.

That does not mean those steps are necessarily wrong, but the path needs to be understood. If nobody can explain which data was used, what changed along the way and why, differences between reported figures become much harder to resolve.

3. How timing and data updates change the figures people see

Sometimes two figures differ simply because they reflect the data at different points in time.

A live production system may update throughout the shift, while a management report uses information only after it has been checked and confirmed. If one person sees the live production total at 15:00 and another looks at the confirmed result the following morning, the figures may differ even though both correctly reflect the information available then.

Shift patterns can create similar differences. A night shift might begin on Monday and finish on Tuesday, with one system assigning production to the calendar date while another records it against the shift start.

Neither approach is automatically wrong, but daily totals will differ unless everyone understands the timing rule.

Delays can also occur before all the data needed for a KPI is complete. Production quantity might be available immediately, while scrap is confirmed later or the reason for a maintenance event is entered at the end of the shift.

That means an important KPI needs clear rules for when the underlying data is updated and when the reported result is considered final.

Without that distinction, teams can end up comparing a live operational figure with a later confirmed result as though they should always match.

4. How KPI definitions and calculation rules change the result

Timing explains some differences, but definitions and calculation rules can create more persistent disagreements.

To compare a KPI consistently, teams need to use the same definition and apply the same calculation.

Overall Equipment Effectiveness, or OEE, is a useful example. Two plants may both report OEE, but the resulting percentage depends on the data and rules used to calculate availability, performance and quality.

Differences can appear in what is treated as planned production time, how certain stops are classified, which ideal cycle time is used and how good or rejected production is recorded.

The same principle applies to measures such as scrap, on-time delivery, output, schedule attainment and unplanned downtime.

If two sites use different definitions or calculation rules, choosing one dashboard figure does not resolve the problem. The organisation first needs to understand why the reported results differ.

That does not mean every local measure needs to be identical. Plants may legitimately track additional information that helps them run their own operation.

However, when figures are being compared across sites or used in group-level reporting, the KPI definition and calculation rules need to be consistent enough for that comparison to mean something.

5. Why local spreadsheets and manual adjustments make consistency harder

Spreadsheets often exist because someone is solving a genuine operational problem.

Perhaps a source system lacks a target needed for reporting, a known data issue must be corrected before a meeting, or information from several systems must be brought together because existing tools do not do it automatically.

Those local fixes can help, but problems start when a spreadsheet becomes an unofficial part of how an important figure is produced, and nobody outside the team understands the logic inside it.

A set of source data might be exported, filtered, corrected, combined with another file and adjusted manually before the final report is produced. If the same process happens every week, that spreadsheet has effectively become another part of the data flow.

The risk is not simply that somebody might make a mistake. It is also that the calculation becomes difficult to explain and repeat.

One plant may apply an adjustment that another does not, formulas may change over time or knowledge of how the file works may sit with one person.

So rather than asking whether spreadsheets should disappear completely, consider the role they play. If the same adjustment happens every reporting cycle, ask whether that rule should be agreed and handled more consistently earlier in the process.

6. Who should own the KPI definitions your teams rely on?

When a reported figure is challenged, it is often unclear who has the authority to explain the definition behind it or decide whether that definition should change.

The IT or data team may manage the system where the information sits, but that does not automatically mean they should decide what downtime, scrap or on-time delivery means for the business.

KPIs need clear ownership of their business definition, while data and technology teams can support how that definition is implemented in systems and reporting.

For an important measure, there should be clear answers to questions such as:

• What does the KPI measure?
• Which events or values are included or excluded?
• Which source data should be used?
• How is the result calculated?
• When is the reported figure considered final?
• Who can approve a change to the definition or calculation?
• Who resolves differences when sites interpret the measure differently?

Several teams may be involved. Production may own what the measure means operationally, the data team may manage how it is calculated, and IT may maintain the systems supplying the data.

Shared responsibility is not a problem as long as those roles are clear. Without it, disagreements tend to return because one team fixes a report while another continues using a different definition or calculation.

7. What a trusted manufacturing data view should look like

A trusted manufacturing view does not require every system to contain the same information.

ERP, MES, SCADA and other operational systems have different jobs, and those differences are useful. The aim is to make the path from the underlying data to an important reported figure clear and repeatable.

People should be able to understand where the data came from, which KPI definition was used, how the calculation was applied, when the information was updated and whether anything was adjusted before the result appeared in the report.

The level of detail can also vary depending on who is using the information.

A plant manager investigating a line problem may need much more detail than a group operations director reviewing performance across ten sites. Those views do not need to be identical, but the figures used for comparison should come from clear definitions and reliable data.

Whatever level of detail people need, they should be able to trace each reported figure back to reliable data, a clear definition and a consistent calculation.

If a number changes, people should understand why. If two reports show different results, it should be possible to trace the difference without rebuilding the calculation from scratch.

8. Which KPI definitions should you standardise first?

Trying to standardise every measure across every plant at once can quickly turn into a large programme that is difficult to prioritise.

A more practical approach is to begin with the measures that regularly create disagreement or affect important operational discussions.

Look for places where several reported figures regularly appear for what should be the same measure; teams spend time checking which result to use, plants use the same KPI name but apply different definitions, manual adjustments happen every reporting cycle, or inconsistent figures make site-to-site comparisons difficult.

Then choose one important KPI and follow its data from the source to the final reported result.

As you trace it, identify where the data originates, which calculation is applied, whether every plant uses the same definition, what is adjusted manually and whether the result changes depending on when the report is produced.

This usually makes the source of the difference much clearer. Sometimes the answer is a common definition; other times, the issue is timing, source data, ownership, or a manual step in the reporting process.

9. How to create more consistent data without replacing every system

Different reported figures do not automatically mean the underlying systems need to be replaced.

In many cases, existing systems may already contain much of the needed information, while inconsistency appears when data is connected, interpreted, or prepared for reporting.

For example, two plants may use different product or machine identifiers even though both systems work perfectly well locally. Rather than replacing either system, the group may need a consistent way to map those identifiers when combining information.

The same applies to KPI calculations. If two plants use different rules for downtime, implementing a new reporting platform will not resolve the disagreement unless you address those rules as well.

The improvement may therefore involve better connections between existing data sources, agreed mappings, clearer calculation rules or removing a repeated manual step.

The key is to address why the reported figures differ rather than assuming the answer is a major technology change.

10. What trusted manufacturing data looks like in practice

Imagine a manufacturing group that wants to compare reported downtime figures across several plants.

Each plant already records production stops, but the monthly review often turns into a discussion about why the reported results don’t match.

When we trace the data and calculations, the differences become clearer. One plant treats planned cleaning differently from another, while a third adjusts several downtime reasons manually in a spreadsheet before producing its final report. The systems also update at different times, so some figures reflect live information while others use confirmed data.

Adding another dashboard would not solve that problem.

Instead, the group agrees on what the shared downtime measure should represent, which events it includes, which source data it uses, how it calculates the result, when it becomes final, and who owns the definition.

Local teams can still track additional measures that are useful for running their individual plants, but the figures used for group comparison are now produced using the same agreed rules.

The existing systems can keep doing their jobs, but the way the data is brought together and turned into a figure that can be compared across the group changes.

As a result, the monthly conversation can shift from “Why do these numbers not match?” to the more useful question: “What is causing downtime to change, and what should we do about it?”

What to do next

If different teams or plants regularly report different figures for what should be the same KPI, choose one important measure and follow the underlying data from its source to the final report.

Look at where the data comes from, when it is recorded, how the KPI is defined, which calculation rules apply, what changes manually, and who owns the definition.

That will usually show whether the difference comes from source data, timing, definitions, calculation rules, manual adjustments, or how information moves between systems.

Once you know the cause, you can focus improvements on that part of the process rather than trying to standardise or replace everything at once.

Want to see the bigger picture?

The guide “Making Better Use of Manufacturing Data” explores six common manufacturing data issues, including conflicting figures, slow reporting, and delayed production information, and what to review first.

-> READ THE GUIDE