Turning automotive data into business insight
As a Data Science Intern at Harris Group, I analyzed automotive inventory, sales and vehicle movement data, built Power BI reporting, and presented findings to the Managing Director and senior leaders. One recommendation, about the data foundations behind the reporting, was approved and implemented.
Original dashboards, data and internal reports are confidential, so they aren't shown. This case study describes the work and what I learned from it.
The business context
Harris Group operates in the automotive industry, where knowing what stock you hold, where it sits and how fast it moves matters to operational performance.
Sales and inventory information lived in several different reporting sources. That made a consistent view of the business harder to get, and it raised practical questions: which stock was moving, which wasn't, how stock compared with demand in each region, and when to replenish.
What I worked on
Power BI reporting
Dashboards and reports giving visibility into stock levels, sales activity and vehicle movement, plus extra reports for specific management questions.
Dead stock
Identified slow-moving automotive parts with no recorded sales for 180 days or more, highlighting inventory that warranted management attention.
Regional allocation
Compared sales activity and stock distribution across regions, including electric and diesel vehicles, to inform discussions about where vehicles should go.
Replenishment and safety stock
Looked at historical usage and supplier lead times to support decisions about stock thresholds and replenishment.
Why this matters commercially: stock that doesn't move ties up working capital and space, while too little stock means missed sales. The aim was to give managers a clearer view of both.
How I approached it
Each piece of work followed roughly the same path, starting from the business question rather than the dashboard:
- Understand the questionFind out what decision the request was really meant to support.
- Gather and prepare the dataPull the relevant information together from the different reporting sources and clean it.
- Model and measureOrganize it into metrics that answer the question.
- Build the viewA Power BI dashboard or report that people could explore.
- Present and follow upWalk leaders through what it shows, and answer the follow-up questions.
An illustration of how I approached the work, not a Harris Group deliverable.
The recommendation that was approved
Working across fragmented sources made every analysis slower and less consistent. So alongside the dashboards, I made a recommendation about the data itself.
What I found
Fragmented reporting
Related sales and inventory information was spread across several sources, which made consistent reporting harder.
What I recommended
Consolidate and remodel
Combine the relevant sources and use a star-schema model: one central table of records linked to lookup tables such as product, region and date.
What happened
Approved and merged
The recommendation was approved, the relevant reporting sources were merged, and the dashboards were then used in leadership reviews.
My part was identifying the problem, making the case for the change and supporting the move to more consistent reporting data. The lesson I took from it: good analytics depends as much on a reliable data foundation as on the visuals built on top.
Working with senior leaders
Much of the internship involved the Managing Director, executives and department heads. That meant:
- turning ad hoc requests into the real business question behind them;
- explaining findings in plain language to non-technical leaders;
- connecting technical recommendations to operational priorities;
- presenting dashboard outputs and responding to follow-up questions.
What I learned
Start with the business question
Understand the operational problem before reaching for a dashboard.
Structure matters
How data is organized decides how consistent and maintainable the reporting can be.
Communication counts as much as technique
Leaders need to know why an insight matters and what decision it informs.
Context changes the reading
Inventory numbers mean more when read against availability, regional demand and working capital.
No figures for savings, revenue or efficiency gains have been independently verified, so this case study doesn't claim any.
Next projectUCD Research: does giving people control change how they feel about a platform?