Spotting differences before they become an inventory problem.

During an inventory, different areas were recorded at different times.

The so-called low zone — the areas of the sales floor accessible to customers — was counted on the actual inventory day.

Stock in the high zones, on the other hand, could already be recorded by the stores in advance.

Depending on the store, this pre-count started at different times. That created a period between recording an article and the actual inventory day in which stock could continue to change through goods movements.

All the more reason to spot anomalies during preparation and to check the recorded quantities for plausibility.

System-side evaluations were already available for checking the pre-count.

While working with the data, however, I noticed a constellation that could make a quick overall assessment more difficult:

An article could have been recorded in several different counting zones. In the existing view, these entries appeared separately from one another.

For a robust assessment, though, it seemed essential to me not only to look at individual recording rows, but to compare the total recorded quantity of an article with the current total stock.

That is where the idea for an additional analysis came from.

For my analysis I used data from SAP and then prepared it with my own Excel macro solution.

The aim was to combine the different pre-counts of an article and then compare them with the current total stock information.

This produced an overall view per article:

A deviation on its own does not yet explain its cause.

That is why it mattered to me not to provide the analysis merely as a list of figures. For different constellations I added notes on how the results could be interpreted and which possible causes could be considered during a further check.

Conspicuous differences could, for example, indicate that goods movements, entries or stock changes should be looked at more closely.

The analysis was not meant to automate a professional decision. It was meant to help users recognise relevant anomalies more quickly and then check them specifically.

Alongside the technical evaluation, I therefore also produced an understandable guide for the stores.

It explained different result situations and linked them with possible ways of checking.

For me that was an important part of the solution: an analysis only delivers its full value once the user understands what the information shown means and which next steps can follow from it.

In this way, the project connected data preparation, professional interpretation and knowledge transfer.

The evaluation was made available to the stores together with explanations.

My aim was not simply to send out yet another Excel file. Recipients should be able to follow how the results are to be read and in which constellations a closer check can make sense.

Depending on a store's situation and development, the analysis could be created again and provided in an updated form.

In this way, the technical evaluation became an instrument that combined analysis, communication and operational support.

The main benefit lay in not looking at anomalies only immediately before inventory day.

By comparing the pre-count with the current total stock, deviations could be made more transparent during inventory preparation and then checked in a targeted way.

As it was used further, my perception at the time was that differences from the pre-count could be reduced, because anomalies became visible and workable earlier.

For me that was a good example of how data analysis can help to accompany a process while it is still taking shape, instead of only checking it at the end.

In this project, the starting point for me was not Excel, but a professional observation.

I wanted to understand why an existing view did not always offer the transparency I wanted for practical work on a particular question.

Only then did the technical solution follow.

That order still matters to me today:

Not first considering which tool you could use — but understanding which problem actually needs solving.

Once the question is clear, data and technology can help to develop a workable solution from it.