How Data Challenged £100,000 of Inventory Assumptions
A question landed on my desk that sounds deceptively simple:
“Are we holding the right amount of stock?”
The business carried thousands of inventory lines sourced from suppliers across multiple countries, each with different lead times, risks and purchasing patterns.
The challenge wasn’t a lack of data.
It was an abundance of it.
Years of stock transactions, purchasing decisions, supplier information and inventory levels existed across multiple systems.
The question was how to turn all of that information into something useful.
The Problem With “Looks About Right”
Historically, many stock decisions had evolved naturally.
Some products had higher stock levels because they’d always had higher stock levels.
Others reflected supplier concerns, past shortages, or decisions made years earlier under different circumstances.
None of this was necessarily wrong.
But it raised an interesting question:
If we were setting these stock levels for the first time today, would we choose the same numbers?
That’s where the project began.
Building the Model
The goal wasn’t to replace experience.
It was to support it.
I built what eventually became an Inventory Intelligence Model.
The approach was deliberately straightforward.
For each stock item, the model analysed:
- Historical annual demand
- Average weekly usage
- Current stock holding
- Existing purchase orders
- Supplier lead times
- Supplier location
- Stock coverage requirements
From there, the model calculated:
- Suggested minimum stock levels
- Suggested maximum stock levels
- Recommended reorder points
Rather than applying a blanket rule, stocking profiles varied depending on supply chain risk.
A supplier on the other side of the world requires a different strategy to one a few hours away.
What the Data Revealed
The interesting part wasn’t the calculations.
It was the conversations.
The model highlighted products where stock holdings appeared significantly higher than usage patterns suggested.
It also identified areas where stock levels were arguably lower than the associated supply chain risk justified.
In other words, it challenged assumptions.
Some recommendations were immediately accepted.
Others prompted useful debate.
And that’s exactly what I wanted.
The objective was never to produce a spreadsheet everyone blindly followed.
The objective was to provide evidence.
The Outcome
When the analysis was complete, the model identified over £100,000 of potential inventory optimisation.
More importantly, it provided a repeatable framework for future decision-making.
Instead of relying solely on instinct or historical practice, stock decisions could now be supported by data.
Not dictated by data.
Supported by it.
What I Learned
One thing this project reinforced is that data rarely changes minds by itself.
People do.
The most valuable part of the work wasn’t the model.
It wasn’t the formulas.
It wasn’t the reporting.
It was creating enough confidence in the information that people were willing to question long-held assumptions.
Good data doesn’t make decisions for us.
It helps us make better ones.
Final Thought
One of the biggest misconceptions about data is that its purpose is to provide answers.
In reality, its greatest value often comes from asking better questions.
The inventory intelligence model didn’t tell people what to do.
It gave them the confidence to challenge assumptions, validate experience and make more informed decisions.
And sometimes, that’s where the biggest improvements are found.