An international retail client came to us with a simple but serious problem:

Their e-commerce business was growing, but they were not able to fulfil around 30% of the orders.

The business had around 400 hypermarkets and supermarkets across the country.

Customers could place orders through the company’s own e-commerce website as well as through other marketplace platforms.

The stores were connected to these platforms and were expected to fulfil customer orders from their existing inventory.

On paper, the model looked simple.

In reality, it was creating a major operational problem.

The same inventory was serving two customers:

the customer standing inside the store and the customer ordering online.

And the fulfilment rate was only around 70%.

That meant roughly 3 out of every 10 orders were facing a fulfilment problem.

The first question was not:

“How do we increase inventory?”

The first question was:

“Where exactly is the problem happening?”

We started with the data.

We looked at four areas.

1. Order fulfilment lead time

We studied how much time the stores actually had to pick and service an online order.

2. Demand forecast

We checked the demand quantity being generated by the application and whether it reflected actual store-level demand.

3. Inventory update frequency

We examined the time gap between the physical inventory available in the store and the inventory quantity visible on the e-commerce platform.

4. Store-level fulfilment

We mapped which stores were servicing which platforms and where orders were getting missed.

That analysis started showing us something important.

The problem was not simply a lack of inventory.

The bigger problem was inventory visibility and inventory accuracy.

Different platforms were generating orders at different times.

The company’s own website was generating orders.

Customers were simultaneously buying from the physical stores.

The store inventory was changing continuously.

But the e-commerce platforms were not receiving inventory updates quickly enough.

So the system could show a product as available online even when the store could no longer fulfil the order.

That created an out-of-stock situation after the customer had already placed the order.

Then we applied Pareto.

We analysed the SKU-level data.

The result was clear.

Around 60–70% of the fulfilment problem was concentrated in the top-selling SKUs.

That changed our approach.

We did not try to fix hundreds or thousands of SKUs at once.

We went after the SKUs that were creating the biggest impact.

We worked with the stores to identify and correct ambiguous inventory quantities.

We looked at the actual stock on the ground.

We compared it with the system stock.

And we focused on improving the connection between the store inventory and the e-commerce platforms.

One important recommendation was technology-related.

The ideal solution was real-time inventory synchronisation.

But implementing a completely real-time system would have required additional technology changes and investment.

So we looked at what was commercially practical.

Our recommendation was:

If real-time synchronisation is not immediately possible, reduce the inventory update cycle to around 10 minutes.

That small change could significantly reduce the gap between:

Physical Stock → System Stock → Online Availability

The technology alone, however, was not enough.

We went to the stores.

This is where Gemba became important.

Data can tell you where the problem is.

But going to the actual workplace helps you understand why it is happening.

We checked what was happening at store level.

We looked at the actual inventory.

We understood how orders were coming from different platforms.

We checked how the store teams were responding.

We compared system inventory with physical inventory.

And we connected those observations back to the data.

That combination gave us a much clearer picture.

Technology showed us the symptom.

Data showed us the concentration.

Pareto showed us where to focus.

Gemba showed us the operational reason.

The result?

The out-of-stock problem reduced from around 30% to approximately 11%.

That took the fulfilment rate from around:

70% → 89%

in a relatively short period.

More importantly, the client did not need to simply keep adding inventory to solve the problem.

We improved the way existing inventory was being seen, updated and allocated for e-commerce fulfilment.

The client subsequently took the process forward and continued following the improvements internally.

What did we actually change?

It was not one big technology project.

It was a combination of several practical interventions:

• Analysed fulfilment lead time

• Studied demand forecasting logic

• Mapped inventory update frequency

• Identified store-level fulfilment gaps

• Created SKU-level Pareto analysis

• Focused on high-impact SKUs

• Checked physical inventory against system inventory

• Identified ambiguous inventory quantities

• Recommended faster inventory synchronisation

• Recommended approximately 10-minute inventory updates where real-time integration was not immediately practical

• Conducted Gemba-level validation

• Connected technology data with store operations

And that is what made the difference.

The bigger lesson for retailers

Many retailers assume that e-commerce fulfilment problems can be solved by:

“Keep more stock.”

I don’t agree.

Sometimes the inventory is already sitting inside the store.

The problem is that the system does not know the right quantity at the right time.

And when that happens, the customer sees:

“Available”

places the order,

and then receives:

“Sorry, this item is unavailable.”

That is not just an inventory problem.

It is a technology + data + process + store execution problem.

For retailers operating an omnichannel model, the real question is not simply:

“How much inventory do we have?”

It is:

“Do we know exactly where that inventory is, whether it is actually available, and how quickly can we make that information visible to the customer?”

That is where e-commerce, store operations, data analytics and Gemba have to work together.

And sometimes, a relatively simple change in the way inventory is updated can create a much bigger impact than simply adding more stock.

From 70% fulfilment to 89%.

Not by throwing more inventory at the problem.

By finding the real problem first.