When is automatic replenishment too early?

When balances are unreliable, periods of scarcity are not marked, promotions are indistinguishable from normal demand or supply terms are often changed manually. In this case, the model can very accurately optimize the wrong story.

Does the forecast have to place an order with the supplier itself?

Not necessarily. The first mature stage is often a recommendation to the planner with a clear reason and a list of exceptions. Automatic approval is only worth turning on for stable goods and locations where the model result has already been verified.

Which indicators best represent the value of the forecast?

It is necessary to assess together the lack of goods, surplus stocks and write-downs or write-downs. The mathematical error of the forecast alone does not indicate whether the decision has improved the real stock result.

How do you include a store that has no sales history yet?

For the first plan, data from a similar store can be used, separately marking this assumption. The size, range and delivery schedule of the new location need to be evaluated, and when actual sales are received, check where the initial comparison was not suitable. The initial replenishment quantities are reviewed by the planner.

Retail chain demand and inventory planning system

How the solution works

Sales history separates stock and commodity shortages periods
Estimated need is comparable to available and incoming quantity
The proposal for a supplement is checked against supply, packaging and storage restrictions
The quantity confirmed by the planner is transferred to the order, and exceptions are saved for viewing

Key challenges

Demand forecasting and addition overstate local demand

Solution capabilities

Demand by product and point of sale

The planner sees the expected need change by location, season, and campaign, marking unreliable data separately.

Assessment of Shortage History

Product-free periods are used to estimate forecast thresholds so that zero sales would not be considered evidence that the item was not needed.

Selection of the amount of addition

The quantity offered is combined with available and ordered stock, delivery time, packaging and storage restrictions.

Risk-based validation

An unusual or under-based recommendation is checked by the planner; approved automatic rules may be applied to cases that have been stable.

Verification of the outcome

The error of the forecast is assessed together with the lack of goods and the excess value of the stock, so one improved metric does not hide the worse service.

Business context

Demand forecasting and addition overstate local demand: In addition offers, there is an inherent demand for a particular store, promotions and periods when the item was not available. Stocks accumulate in some stores, and goods are missing in others; transfers and urgent additions increase the cost of service.
Each store is catered to its buyers: A uniform replenishment solution does not take into account the different local demand. The goods and point-of-sale plan helps to send supplies where they have a greater chance of being bought. This allows for better accessibility without growing all network stocks at the same rate.

Core features

Demand by product and point of sale
Assessment of Shortage History
Selection of the amount of addition
Risk-based validation
Verification of the outcome

Key integrations

POS / Sales Data: SKU and location sales, returns and shortage period signals.
ERP / WMS: stock, ordered stock, terms of supply and warehouse restrictions.
Cross-channel pricing and promotion management system: Stock calendar and price changes that alter the interpretation of demand history.
Supply Planning Data: Supplier terms, minimum order quantities and other replenishment restrictions.

Potential impact (%)

The ranges indicate an illustrative relative change in the metric under the stated assumptions. Results depend on the starting position and actual use of the solution. Percentages for different metrics must not be added together.

Share of goods shortage cases in active range

2–12%Decreasing

This illustrative scenario assumes that 10-30% of the metric is attributable to addressable planning and execution shortcomings. That share is assumed to fall by 20-40%. Company data is needed to verify both the addressable share and the resulting change.

SkU and location days are counted without selling stock, separately marking supply disruptions that planning could not avoid. The percentage is calculated on the number of all inspected goods and location days.

Share of surplus stock in selected category

2–12%Decreasing

This illustrative scenario assumes that 10-30% of the metric is attributable to addressable planning and execution shortcomings. That share is assumed to fall by 20-40%. Company data is needed to verify both the addressable share and the resulting change.

The part of the stock above the agreed limit for days or turnover is valued, for the same set of category and locations. The percentage is calculated from the value of the total stock in the category being checked.

Sales revenue of the trading network

1–6%Increasing

In the example scenario, 20-40% of the original indicator value is associated with a change in stock allocation made in the trading network. This share is predicted to grow by 5-15% without other conditions changing.

After a change in stock allocation has been effected, the network's sales revenue is compared, along with monitoring the value of stocks, prices, depreciations and shortages of goods.

Conditional calculation scenarios. The assumptions have not been validated against client measurements.

When this solution is relevant

Forecasting and replenishment rules do not always distinguish between promotions, disadvantages, season, local events, delivery deadlines and the actual state of the shelf
Some stores accumulate surpluses, others lose sales, and employees adjust system offerings manually
The addition does not respond adequately to local demand and exceptions
Customers are promised stock that is already reserved or unavailable for sale

Implementation requirements

The addition forecasts distinguish between low demand and sales lost due to lack of goods. Historical estimation is based on data available at the time of decision and actual delivery terms, and automatic order limits are determined by the reliability of the forecast and stock risk.

Further development options

Recommendations automatic validation based on reliability
Comparison of supply disruption scenarios