Management team watching the KPI dashboard on a large screen in the meeting room

Oct 19, 2026

AI Isn't There to Write Reports. It's There to Make Commercial Decisions.

From the dashboard era to the era of decision systems: where will the real revolution in retail happen?

Strategic Management

Retail

Context

What grows a company is not owning technology; it is being able to turn technology into the right commercial decisions.

Imagine a Monday morning at a retail company.

The sales manager is reviewing last week's turnover.

The category manager is preparing a report on unsold products.

The purchasing department is calculating stock requirements.

The marketing team is evaluating campaign results.

The finance department is trying to complete its profitability analyses.

Everyone is dealing with data.

Everyone is preparing reports.

Everyone is measuring something.

But is the company really making better decisions?

This is exactly the point I question.

Because for years, most companies invested in making data visible.

ERP systems were installed.

Reporting dashboards were built.

Business Intelligence applications were developed.

But in many organisations, a serious gap opened up between the speed of producing data and the speed of making decisions.

Companies began to see more. But they could not turn everything they saw into commercial action.

I see the real potential of AI in its ability to solve this problem.

Analysis

1. A dashboard tells us what happened. A decision system questions what should be done.

A traditional reporting system can provide this information:

“Sales of product X fell 18% compared with last week.”

This is valuable information.

But on its own it is not enough.

Because the questions the manager needs to answer remain.

Why did sales fall?

Did customer traffic decline in the stores where the product is sold?

Was the product positioned in the wrong stores?

Is the stock in the right locations?

Did competitors change their prices?

Has overall demand in the product's category declined?

Was marketing support insufficient?

This is where the difference of AI-assisted decision systems can emerge.

These systems can relate separate data sets to analyse possible causes and generate alternatives for commercial action.

But an important distinction must be made.

AI making a recommendation does not mean the recommendation is automatically correct.

In well-designed systems, the data, assumptions and expected effects behind a recommendation should also be visible.

That way, the manager is confronted not only with the result but also with the logic of the decision.

2. Retail's invisible problem: the right product in the wrong place

Imagine a company has 10,000 units of a product in its warehouse.

At first glance, it may seem there is enough stock.

But the real issue is not the total quantity of stock.

It is in which store, at what time and under what demand conditions the product is located.

A product that sells fast in Istanbul cannot be expected to perform the same way in another city.

The customer profile in a shopping mall may differ from that of a high-street store.

Demand in tourist locations may also differ from that of stores serving a settled customer base.

So distributing the same product to all stores in the same quantity may be operationally easy, but it is not always commercially right.

AI-assisted demand forecasting and stock optimisation can evaluate store-level sales history, seasonality, the campaign calendar, availability and other relevant variables together.

The aim here is not to leave every decision to the algorithm.

The aim is to notice relationships the manager might miss — earlier.

Because in retail, stock sitting in the wrong place does not only create storage costs.

It can also cause the loss of a sale that could have happened in another store.

Inventory stock count by scanning barcodes in an orderly store warehouse

A planner doing sales and operations planning with a tablet in the warehouse

In Depth

3. Category management is being redefined with AI

McKinsey's assessment published in April 2026 states that AI agents could significantly reduce the time retail category managers spend on repetitive work, potentially freeing up to 40% of their time for more strategic activities.

To me, what matters here is not only the time saved.

It is the change in the category manager's role.

In the traditional structure, a category manager may spend a large part of their time reviewing reports.

Which product is selling?

Which one is moving slowly?

Which store has excess stock?

Which product's price should change?

In the new era, a significant part of these analyses can be carried out with AI support.

This changes the topics the category manager can devote more time to.

Portfolio architecture.

New product opportunities.

Changes in customer needs.

Differentiation strategies.

Supplier relationships.

Category profitability.

In other words, category management can move from a task that reports on the past towards a commercial responsibility that shapes the future.

4. Stock turnover should no longer be seen only as an operational KPI

In retail, inventory management is often treated as a logistics or purchasing matter.

Yet stock is also a financial decision.

Because every product waiting in stock ties up part of the company's capital.

Slow-moving products can reduce financial flexibility.

Insufficient stock, on the other hand, can lead to lost sales.

That is why an AI-assisted commercial system should not look only at sales velocity.

It should evaluate indicators such as days of stock, gross margin, contribution margin, reorder lead time, product life cycle and store-level demand together.

For example, a product with high sales volume may not always be the best stock investment.

If the product's gross margin is low, its supply time long or its demand very volatile, the financial outcome may differ.

Similarly, a product with lower sales volume may be strategically important in terms of the contribution margin and capital efficiency it provides.

That is why we cannot say the right commercial question is only:

“Which product sells faster?”

The real question should be:

“Which product gives us the best commercial return on the capital we tie up in it?”

5. Production and retail should be part of the same decision system

One of the most important application areas of AI transformation is getting production and sales organisations to work together.

Data integrity becomes far more important, especially in companies that run their own production while also operating retail, wholesale, contract manufacturing or private label businesses.

Consider a cosmetics manufacturer.

Demand from sales channels differs.

The stock needs of retail stores differ.

The order cycles of wholesale customers differ.

The product specifications and delivery commitments of private label customers differ.

The production planning department has to take all of these demands into account.

But if this data is kept in different systems at different levels of freshness, making the right production decision becomes difficult.

This is where an AI-assisted ERP approach can offer a new opportunity.

The system can compare validated sales forecasts with existing stock.

It can calculate raw material requirements according to bills of materials.

It can evaluate supply lead times.

It can identify constraints in production capacity.

It can create alternative production scenarios.

But human control may be needed to approve the production plan.

Because some decisions are not based on numerical data alone.

Supplier reliability, strategic customer relationships, product quality risks and production priorities also matter.

That is why my goal in operations management with AI is not systems that remove human decision-making altogether, but systems that enable people to make better decisions.

Roadmap

6. AI's biggest risk is accelerating the wrong decision

While describing the technology's potential, we must not ignore the risks.

If an AI system works with faulty stock data, it can generate wrong order recommendations.

If sales data has not been cleaned of campaign effects, it can misinterpret demand forecasts.

If return data is missing, it can create a profitability picture that does not actually exist.

So data quality is one of the fundamental conditions of AI investments.

In addition, authority limits must be clearly defined.

A system can recommend price changes to managers.

But it may not be right for it to apply every price change automatically.

A system can calculate reorder quantities.

But there should be an approval mechanism for purchasing decisions above a certain amount.

A system can recommend transfers between stores.

But the logistics cost and operational constraints of the transfer must also be evaluated.

That is why I believe three core principles must be preserved in commercial AI architectures:

Data accuracy, explainability of decisions and authorised execution.

Without these three elements, automation can bring speed but also amplify the impact of errors.

Quality specialist inspecting amber bottles in a cosmetic manufacturing facility, private label production

Conclusion

7. Where should a commercial AI transformation start?

I do not think companies must start their AI transformation with a large technology investment.

First, the company's most important commercial bottleneck should be identified.

In one company, the problem may be stock turnover.

In another, store productivity.

In another, purchase planning.

And in another, customer churn.

In the first stage, it may be more sensible to choose a single, measurable problem.

For example, a pilot system that generates reorder recommendations in a limited number of stores can be set up.

The current method can be compared with AI-assisted recommendations.

The stock-out rate, days of stock, lost sales and contribution margin can be tracked.

Once the results have been evaluated reliably, the scope of the system can be expanded.

This approach allows commercial risk to be controlled while investing in technology.

It also helps the organisation adapt to the new way of working.

8. The manager of the future won't read more reports. They will make more right decisions.

I do not think the role of managers will disappear in the future of retail management.

But I do think the competencies expected of managers will change.

Data literacy will become more important.

Being able to evaluate commercial scenarios will become more important.

Being able to question the recommendations of AI systems will become a new management skill.

Ensuring that different departments can work from the same data will become critical.

Alongside this, people management, customer experience, brand intuition and strategic judgement will keep their value.

Because although algorithms can develop powerful analyses based on past behaviour, every new commercial opportunity is not simply a repetition of past data.

Creating a new category.

Redefining a brand's position.

Noticing a customer need that has not yet been clearly expressed.

These still require strong commercial leadership.

Final word

For years, we told companies about the importance of data-driven management.

Today we need to talk about the next stage.

Just seeing data is not enough.

Making sense of data is not enough.

We need to turn that data into the right commercial decisions and implement those decisions in the field.

In my view, this is where the real value of AI in retail will emerge.

The ERP can be the company's memory.

The dashboard can be the company's mirror.

AI, when designed correctly, can strengthen the company's decision-making mechanism.

But what grows a company is not owning technology; it is being able to turn technology into the right commercial decisions.

Sources

McKinsey – From Dashboards to Decisions