When Will Our Dead Stock Sell? — The Real Problem Is How You Order
When Will Our Dead Stock Sell?
Few challenges in brand operations come up as repeatedly as inventory problems.
When a new product doesn't sell as well as expected, or stock piles up as a season winds down, the mood in the meeting room quickly turns heavy.
So we keep coming back to questions like these.
"When will this inventory finally clear out?" "Did we go too aggressive this time?" "Was our forecast wrong?"
But all of these questions look at the outcome.
Inventory is the result of sales and ordering, not the cause of the problem.
The core reason inventory piles up is actually decided much further upstream.
There's just one thing we want to say in this article.
The essence of the inventory problem isn't sales — it's 'how you order.'
Dead stock isn't created when a product starts selling slowly; it's created the moment you hit the order button.
So we need to change the question.
"What is it about the way we order that makes the same inventory pattern repeat every season?"
This article walks through the answer to that question
data → forecasting → ordering structure → automation
across these four stages.
1. When data doesn't flow, ordering wobbles
Brands say, "We make decisions based on data."
But when you actually break down the ordering process,
the data is mostly scattered across many places.
Sales data differs by platform,
inventory is checked in the ERP,
ad data lives in the marketer's report,
lead times and MOQs are in the manager's head,
and the promotion schedule is shared separately.
Pulling all of this together right before ordering and making a comprehensive judgment is nearly impossible.
So in reality, ordering happens like this.
"Looking at the recent sales trend…" "Last season we moved about this much…" "Just in case, a little more…"
In other words, data is merely a reference, and ordering ultimately comes down to instinct, experience, and the mood of the room.
This is exactly where the problem begins.
■ When data is scattered, judgment can't be consistent
Some days you go aggressive, other days conservative; for some SKUs past experience dominates, for others you simply rely on the trend.
For a brand with hundreds or thousands of SKUs, ordering effectively becomes a decision that changes from person to person, every time.
This is where the seeds of dead stock are planted.
So the first thing you need isn't to gather more data —
it's to make all the data needed for ordering decisions visible on 'one screen.'
Without this, every later stage will wobble.
2. Forecasting is a starting point, not a solution to ordering
Many brands say, "We've built a sales forecasting model."
And naturally, they expect this:
"If the forecast gets more accurate, ordering will stabilize."
But in reality, you hear this more often:
"The forecast is pretty accurate, yet we ended up with leftover stock again." "The stockout risk is still high." "Forecasting and ordering feel like they're running on separate tracks."
Why does this happen?
The reason is simple.
■ Forecasting is a 'number'; ordering is a 'judgment'
Even if the forecast gives you the number "130 units will sell next month," reality comes with constraints like these.
MOQ of 200 units
Packing unit of 120
30-day lead time
Limited warehouse space
Budget limits
Category strategy
The real difficulty here isn't the number — it's the judgment.
Do you order 200 of a SKU that's expected to sell 130?
Do you place the order this month to account for lead time?
Where should you prioritize allocating budget?
How do you balance things across categories?
So no matter how good the forecast gets, without an ordering structure, the order outcome still depends on instinct.
In other words,
forecasting is a necessary condition for ordering, but never a sufficient one.
Between forecasting and ordering, you absolutely need a "structure."
3. Ordering should be run as an 'optimization structure,' not a model
Many people think of ordering as a simple calculation problem.
But ordering is, in effect, a constraint-based optimization problem.
Forecasting is just one element of it.
Here are examples of the factors you need to consider when deciding an order.
Sales forecast
MOQ
Packing unit
Lead time
Channel-by-channel sell-through rate
Inventory cost
Supply stability
Season end date
Cannibalization
Logistics processing limits
Only by considering all of these variables at once do you arrive at the order quantity that answers "how many of this SKU is it most reasonable to order right now".
But no human can calculate all of these factors weekly, daily, across hundreds of SKUs.
So in reality, this is what happens.
Order in bulk if it seems like it'll sell well,
order "just one box" if it's uncertain,
keep piling volume onto only the SKUs that sell well,
let similar SKUs cannibalize each other,
and place blanket orders that ignore channel-by-channel speed differences.
Repeat this just a few times and the inventory structure quickly becomes unstable, with dead stock soon claiming a corner.
In the end, what's needed isn't for people to calculate harder —
it's to structure the brand's ordering principles, and to have the optimal order quantity calculated automatically within that structure.
Order optimization isn't "a system where AI makes the decision for you" — it's closer to a framework that applies the brand's strategy consistently.
4. The faster the judgment, the more valuable it is — which is why you need automation
Even with the structure well in place, operations always change faster than forecasts.
Sudden spikes in sales,
unexpected supply delays,
abrupt shifts in channel speed,
changes in competitors' pricing strategies,
peak-season/event effects.
If your judgment is slow at these moments, inventory starts going wrong again. That's why you need automation.
But don't misunderstand here.
Automation ≠ a system that places orders for you automatically.
The key is maintaining the "flow of judgment."
When the forecast updates,
when constraints change,
when sell-through rates shift,
→ the ordering structure recalculates automatically, helping people not miss anything.
Order automation is a mechanism that keeps the judgment structure unbroken, in step with operational speed.
More important than accuracy is speed and consistency of judgment.
5. So where should you start changing things?
What can you do right now?
1) Write down which materials you're looking at in ordering meetings
Just jotting down which files, which data, and which questions drive your decisions immediately reveals how scattered your brand's ordering structure really is.
2) Turn recurring sentences from meetings into policy
"It's fine if a little of this SKU is left over." "This line must never go out of stock."
Statements like these are already the prototype of your brand's operating policy.
Documenting them as official policy gives your ordering a more consistent decision structure.
3) Then comes AI and automation
Introduce AI without structure and you get "a system that's wrong a little more precisely."
Structure → policy → integration → forecasting → optimization → automation
This order has to be respected.
6. The final question: when will dead stock sell?
Honestly, no one can give a definitive answer.
Too many external factors and chance events are tangled up in it.
But the more important question is this.
If we keep ordering this way, won't the same inventory pattern repeat next year too?
Inventory is the output value of your operating structure.
If you want to change the output, you have to redesign the structure first.
Rather than hoping dead stock eventually sells, build an ordering structure that doesn't produce dead stock in the first place.
This is the most realistic and sustainable way to solve the inventory problem.
If you'd like to review your current ordering approach
If, while reading this article,
you thought "this sounds exactly like us…"
it may be a sign that your current ordering structure has already hit its limits.
Dalpha connects sales data, inventory, lead times, MOQs, and channel-by-channel sell-through rates into a single structure
to build an AI solution that designs each brand's ordering decision structure and makes that structure actually work in real operations.
Where your brand's ordering approach is currently wobbling,
how judgment is being distorted after the forecast,
where dead stock is being structurally created —
these are the points we work through with you from a data and structural perspective.
Before "adopting AI," if you'd like to first check whether your current ordering approach is truly reasonable, feel free to have a casual chat with us.
Minhyuk Choi
