Hello.

When running a store or a company, anyone experiences this situation at least once.

• "This time I ordered too little and the popular item sold out quickly..."

• "Next time I need to order more!"

• "This time I ordered plenty, but now I have a lot stacked in the warehouse..."

In the end, orders go between

"Order less this time, order more next time"

like this quite often.

That's why what is often discussed these days is
using AI for demand forecasting and automatic ordering.

Although it's not a feature that's immediately ready to launch,
it's also a planned and reviewed feature that we intend to enhance in the future,
so I would like to introduce what it is and how it can help in practice in advance.

1. Why is 'demand forecasting' important?

The benefits of good demand forecasting are very simple.

1) It can reduce stockouts

• If popular products frequently become "out of stock"
  → It’s like missing out on sales opportunities.

• From the perspective of regular customers,
  → This creates the perception that "There’s nothing I’m looking for every time I come here."

2) It can reduce excess stock

• If you stockpile a lot of products that don’t sell well just in case,
  → It combines warehouse space + cash.

• Especially for perishable goods, it can lead directly to losses.

3) It reduces ordering stress

• It reduces the time spent every month and week pondering, "How much should I order this time?"

• It’s much more relaxing because you can make decisions based on "data-backed numbers".

In short, demand forecasting is
a process of increasing sales while simultaneously reducing inventory risks and stress.

2. What does AI demand forecasting look at to make judgments?

AI demand forecasting may sound grand, but ultimately it’s about reading patterns from data.
Typically, it looks at these types of information.

1) Past sales trends

• How much of this product was sold in the recent weeks/months

• Whether it sells better on certain days (e.g., concentrated sales on weekends, products that spike on weekdays during lunch)

2) Seasonal and event impacts

• How sales volumes change during summer/winter, vacations/semesters, holidays/long weekends

• Whether sales spike during Christmas, year-end, anniversaries, or promotional periods

3) Product characteristics

• Whether it’s a repeat purchase/disposable item (e.g., drinks, snacks, necessities)

• Whether it’s a strongly seasonal product (e.g., iced drinks, winter snacks, limited-time merchandise)

By combining this data, AI can, for example, tell you:

• "In the next two weeks, this product is expected to sell at least 40 units."

• "With only 10 units in current stock, there is a high risk of stockout."

This is the basic work that AI demand forecasting does.

3. How does 'automatic ordering' connect here?

The next step after demand forecasting is automatic ordering (Auto Replenishment).

1) AI calculates the "needed quantity"

• Based on past sales, seasons, events, etc.,
  → it calculates the predicted sales volume.

• For example: "It is expected that about 50 units of this product will be sold in the next 14 days."

2) Compare with current stock

• If current stock is 15 units,
  → the estimated shortage would be about 35 units.

• If there are already ordered quantities,
  → those quantities are considered as well.

3) Order suggestions or automatic ordering

• It shows a draft order suggestion asking, "Would you like to order 35 units of this product?"

• Depending on predefined criteria (minimum/maximum stock, whether automatic ordering is allowed, etc.),
  → it can also create an automatic order request.

we are planning to consider a structure that can gradually introduce this flow:

1. the step where AI suggests order quantities

2. the step where the owner confirms and approves

3. the step where only selected items are converted to automatic orders

4. What if AI demand forecasting is added before going live? (planned direction)

Once AI demand forecasting functionality is added our serveice,
you can expect something like this.

Example 1) Automatic recommendation of "items expected to be short"

• Providing a list of items expected to have stock shortages based on the next 14 days

• For each product

  • predicted sales volume

  • current stock

  • anticipated shortage amount
    will be shown together.

• Examples of buttons below

  • Create order draft

  • Remind me later

Example 2) Notification of "over-ordering risk items"

• Indicating items with too much stock compared to recent sales speed

• For example: "Based on current sales trends, it is expected that it will take more than 90 days to deplete stock."

• Upon seeing this notification,
  you can

  • reduce order quantities

  • consider display positions/promotions/bundle discounts.

Example 3) Seasonal ordering guide

• "In last year's November-December, this product sold more than twice as much as usual."

• "It is expected that this year will follow a similar pattern, so
  it is recommended to adjust order quantities in advance before that period."

Through this kind of guidance,
to assist in preparing orders before the season arrives is also a goal.

5. Features are still ‘in preparation’, what preparations can be made now?

AI demand forecasting & automatic ordering functionality is
not yet actually implemented in STOCK 5 MINS AGO,
and we are still internally pondering how to provide it to be the most helpful in practice.

However,
there are preparations you can make now that will be beneficial when utilizing these features properly later.

1) Organize product information

• For the same product,

  • reduce instances where names differ

  • or where option expressions vary

• If you unify product names, option names, and categories,
  → it will be much easier for AI to read patterns.

2) Accumulate stock records centered on going live

• Handle tasks such as receiving, shipping, and stock adjustments
  → If possible, process it at the going live to leave a **log (record)**

• The more consistently and steadily the data accumulates,
  → the higher the accuracy of demand forecasting will be.

3) Leave seasonal and event notes

• For specific periods,

  • year-end sales

  • 1+1 events

  • holiday promotional events
    keeping simple notes will be very helpful later when AI interprets, "There was an event impact at that time."

6. Conclusion: Orders are still decided by humans, while AI assists from the side

AI demand forecasting & automatic ordering
is not about the concept of "Let’s leave all orders to AI from now on,"

but rather a tool that combines the owner's intuition + AI's data analysis
to help make better decisions.

It's much closer to that.

• we will focus on establishing a strong foundation of stable inventory and transaction records

• and we are aiming to gradually implement AI demand forecasting, automatic order suggestion functionality.

For now,
please consider it as a period to "accumulate data well."

If you consistently record the flow of products and inventory in STOCK 5 MINS AGO,
soon AI will be able to

• "Try ordering this much this time."

• "It would be advisable to reduce orders for this product for the time being."

We believe that day will come.

Please look forward to the AI demand forecasting & automatic ordering functionality that we will prepare in the future.
We will continue to think and create to help business owners not solely rely on intuition but utilize data and AI for more reliable inventory management.

🙌