Four regions, one dashboard leadership opened.

Slow, and nobody could say exactly why.

The operation moved analytical instruments and consumables between four regions through more than fifty vendors, logistics partners and customs teams. Delivery turnaround was slow. That much everyone agreed on. What nobody could do was point to a cause, because the evidence was spread across SAP extracts, vendor correspondence and regional spreadsheets that did not agree with each other.

Leadership received a status narrative each month rather than a measure. Escalations arrived as surprises. When a region slipped, the first hour of any conversation was spent establishing whose numbers were right, and the decision came after that, if it came at all.

My brief was to redesign the cross-functional shipment process, and the first thing a process redesign needs is a shared, trusted picture of how the process is actually behaving.

What eighteen cycles showed.

I ran the root-cause analysis as a monthly discipline rather than a one-off workshop. Every late shipment in every cycle was coded to a cause and a vendor, and the coding was reviewed with the regional teams so the categories meant the same thing in Brazil as they did in Germany.

The pattern was not evenly spread. A short list of causes and a short list of vendors accounted for most of the delay, and the mix differed by region. That finding shaped everything that followed: the standard operating procedures were written for the causes that dominated, and the vendor escalation protocol was written for the vendors that kept appearing.

Four measures, one page, exceptions first.

I built the KPI dashboards in Power BI and Excel over SAP data, covering 120-plus active SKUs. The choice of measures was the design decision that mattered most, and I kept it to four.

• On-time deliveryThe outcome the customer feels. Tracked by region against a single target so a lagging region was visible at a glance, never buried in an average.
• Order accuracyThe measure that separates fast from right. A turnaround gain that came with more wrong shipments would not have been a gain.
• Backlog trendOpen orders by age. Improvement in turnaround shows up here first, and so does improvement that is being borrowed from next month.
• Vendor performanceOne row per vendor. This is where accountability lived, and it was the page the escalation protocol was written against.

Three rules governed the layout. Everything fit on one page, because a dashboard that needs scrolling gets summarised in an email instead of opened. Regions were shown side by side at the same scale, so nobody had to ask which line was which. And exceptions came first: the measures that were off target sat at the top, in the same position every month, so a reader who gave the page thirty seconds still left with the right three things.

Budget tracking for the six concurrent capital-import work streams ran on the same cadence, with one column that earned its place: variance against plan to date. It surfaced cost problems while they were still cost problems, before they became timeline problems.

The page leadership actually opened.

A dashboard changes an operation only if it changes what people do on a Monday. Three things made this one stick.

The standardised SOPs and the vendor escalation protocol were written against the dashboard's own definitions, so a breach on the vendor page triggered a defined action rather than a debate. The regional teams reviewed their own numbers before leadership saw them, which turned the monthly review from an audit into a conversation about what to do next. And the numbers were the same numbers every time, from the same extract, which ended the first hour of every conversation being spent on whose figures were right.

Coordinating fifty-plus vendors, logistics partners and customs teams across four regions, the dashboard was less a report than the shared object everyone argued over. That was the point.

What changed, on the record.

Each figure above is on my résumé. Nothing on this page is rounded up.

What I would do differently.

Start with the vendor page. I built it last, after the regional views, and it was the page that changed behaviour most. Put backlog age on the front page from the first cycle, not after the first time a turnaround gain turned out to be borrowed. And automate the SAP extract earlier: every manual step between the system and the page is a place where two people can end up with two versions of the truth, which was the problem I had been hired to remove.

What is real, and what is made.

The original dashboard is Agilent's. What sits beside this case study is a rebuild in public: the same four measures, the same four regions, the same eighteen-month cadence, the same targets, and an intervention in the seventh month with a response curve after it. Every vendor, order, cause and monthly figure on it is generated by a seeded script, tuned so the headline outcomes land where the record says they did.

Treat the exhibit as a demonstration of how I design a dashboard, and the figures on this page as the record of what one did.