Home / Case studies

Real problems, real recovery

Two projects from my work as KPI specialist for an outsourced customer-service operation supporting a large international brand. Organization, exact figures and identifying details are anonymized. The method, the dashboard structure and the approach are what I built.

Case study 01

Finding the hidden driver behind a declining resolution-time KPI

An outsourced customer-service operation delivering support for a large international brand across 12 markets, under strict IT and data-governance rules. All reporting ran on Excel and Access data feeding a Power BI semantic model, and no data could leave the environment.

12: markets affected
2 months: decline window
8: driver KPIs tracked
52 wks: rolling correlation window

The challenge

The operation's core efficiency KPI, Case Resolution Time (CRT%), is tracked separately for escalated and non-escalated cases. It had slid for two months in a row, unevenly. The Premium segment was down in about four markets but still in a neutral band, below the bonus threshold and not yet in penalty territory. The Standard segment was down in roughly eight markets and had fallen deep into penalty territory. Leadership needed to know which of dozens of tracked metrics were actually driving the decline before committing resources to a fix.

What I built

A diagnostic dashboard in Power BI, connected to the operation's Fabric-based semantic model. Its core is a rolling 52-week correlation engine that compares CRT% against about eight candidate driver KPIs, weighted by case volume per market so that noisy low-volume markets cannot drown out the big ones. The output is a matrix with driver KPIs on one axis and market and segment on the other, colored as a heatmap. Behind each cell, a DAX measure returns more than a coefficient. It returns a plain-language verdict: "No effective correlation", "Watch" or "Critical: urgent action". A non-technical manager can read the matrix in seconds. Slicers for time, market, line of business and segment let each manager drill into their own scope.

The outcome

Instead of cross-referencing dozens of KPIs by hand, market by market, managers saw a short list of driver and market combinations marked Critical. A wide, ambiguous investigation became a prioritized action list, and this dashboard became the entry point for the root-cause work in the second case.

Case study 02

From root cause to a 10-week recovery plan

Same operation, same KPI. After the diagnostic work, I was assigned as the internal specialist to own CRT% recovery.

11 of 12: markets with a dedicated action plan
10 weeks: glidepath length
+2 pts: CRT% recovered on plan
DMAIC: project method

The challenge

Diagnosis was not enough. Leadership needed a credible, documented plan to reverse the decline, across markets and segments with different underlying causes. That meant getting buy-in and operational detail from several Operational Managers who did not report to me, finding true root causes rather than symptoms, and turning the findings into a plan the business could be held to.

What I built

I ran the project under DMAIC (Define, Measure, Analyze, Improve, Control) with documentation at every stage. Together with the Operational Managers I planned and ran a structured case-scrubbing exercise, a manual review of individual case records, to find the real root causes market by market instead of relying on aggregate trends. That produced segment- and market-specific action plans for 11 of the 12 markets. Each plan was tied to a modeled glidepath, a week-by-week projection of CRT% recovery. I first built the glidepath in Excel to work through the logic with stakeholders, then rebuilt it in Power BI as a live dashboard comparing actual performance with the plan.

The outcome

Leadership got a documented recovery plan instead of an open-ended "we are working on it", plus a live dashboard showing weekly progress against target. Over the 10-week glidepath, the KPI recovered by roughly 2 percentage points, a meaningful move for a metric that had been sliding for two months. The dashboard also showed early which markets were on plan and which needed a second look.

Stuck KPI, messy data model, no framework?

Let's talk through what is actually going on. The first call is free and there is no obligation.

Book a free 20-minute call Or write an email