Imagine a reporting meeting where three businesses have submitted their production figures. At one site, "complete" means an item has been made. At another, it means the item has passed inspection.
Both figures could be correct. Put them beside each other in a chart, though, and you could mistake a difference in definitions for a difference in performance.
The work before the dashboard
Our ongoing work for a private-equity manufacturing group brings daily production numbers, order books, payroll, and inspections into one operations platform across three companies.
Shop-floor records feed a shared reporting view. The platform uses Power Apps, Azure SQL, and Power BI. Group dashboards refresh twice a day.
For a platform like this, the form someone fills in deserves as much attention as the dashboard a manager reads. If a field is unclear, a shared database can collect inconsistent answers just as easily as a spreadsheet.
A useful reporting system starts with the person who records the work.
One record, more than one use
A production figure needs context: which site recorded it, which date it covers, and what counts towards the total. With those definitions agreed, the same record can support a site manager's review and a group comparison.
Using the same record also removes a reason to copy figures between reports. If someone corrects an entry, the reporting process needs an agreed way to handle that correction.
That raises a decision a chart designer cannot make alone: should yesterday's report change when a site corrects yesterday's figures? The answer depends on how the business uses that report. Someone needs to agree the rule before the software can apply it.
Fresh enough for the decision
A dashboard refreshed twice a day serves a different purpose from a live machine display. The dashboard supports regular business reviews. The machine display may need to show a problem immediately.
Refreshing a chart every minute will not help if its source records only change once a day. A visible "last updated" time helps readers judge whether the figures are current enough for their decision. Missing entries need to be visible too.
What one number can tell you
You can learn a lot about a reporting problem by tracing one figure from the final report to the person who first records it.
If each team changes the figure before passing it on, the project may need an agreed definition before it needs a new dashboard. If everyone agrees what it means but keeps typing it into different files, connecting those records may be the more useful first change.
The project described here is ongoing and anonymous. Read the scope and platform details in our manufacturing case study.