Forecast Repeat Orders by SKU Count

Why SKU-level repeat order forecasting gets harder as you add variants

A brand can be growing overall and still make poor replenishment decisions.

That usually happens when the team looks at total sales, total units, or monthly revenue, but not at the more operational question: which specific SKUs actually repeat, at what rate, and how many units should we buy again?

This gets harder as SKU count rises.

A product line with 3 SKUs can often be managed by feel. A line with 30 variants across colours, sizes, bundles, or regional packs usually cannot. Some SKUs repeat every week. Some only move when promoted. Some look healthy because launch stock sold in one burst, then reorder slowly.

If you replenish from broad sales totals, you tend to:

• overbuy long-tail variants
• underbuy core repeat SKUs
• order to supplier MOQ instead of actual demand
• tie up cash in cartons that sit too long
• create warehouse complexity with stock that does not earn its space

The goal of a good repeat order forecast is not perfect prediction. It is a reliable, repeatable way to turn SKU-level sales and stock data into order quantities by SKU.

What “forecast repeat orders by SKU count” actually means

To forecast repeat orders by SKU count means estimating future replenishment needs at the individual SKU level, based on actual repeat demand rather than broad sales averages.

That is different from three related but separate tasks:

• Total demand forecasting: estimating overall unit sales or revenue across the business
• Repeat order forecasting: estimating which products will continue to sell after launch, promo, or one-off spikes
• Reorder planning by SKU: deciding how many units of each SKU to buy now, given stock on hand, lead times, MOQs, and pack constraints

This distinction matters because supplier orders are placed by SKU, not by headline revenue.

A monthly sales number does not tell you whether black / medium / 12-pack should be reordered, whether coral / small should be reduced, or whether a slow seasonal variant should be left to run down.

For most product brands, useful SKU-level demand forecasting answers five questions:

1. Which SKUs show genuine repeat demand?
2. How fast does each SKU sell in normal conditions?
3. How much demand will occur during lead time?
4. What safety stock is needed to avoid stockouts?
5. What final order quantity is actually possible once MOQ, pack size, and case quantities are applied?

Step 1: Start with clean SKU-level sales and inventory data

Good repeat order forecasting starts with clean inputs. If your SKU list is inconsistent, variants are merged together, or pack sizes are unclear, the forecast becomes unreliable very quickly.

This is why clean SKU structure and master data matter. If your SKU naming still needs work, start with How to Structure SKUs Properly (Before You Print Anything). If your planning fields are incomplete, Product Master Data Sheet: Fields Every Brand Should Include is the next place to tighten things up.

Minimum data fields you need

For each active SKU, you should have:

• SKU code
• product name and variant details
• status: active, launch, seasonal, discontinued, or clearance
• current sellable stock on hand
• inbound stock already on order
• average weekly sales or recent sell-through rate
• lead time in weeks
• safety stock target in weeks or units
• supplier MOQ
• pack size or case multiple
• carton quantity if relevant
• primary supplier
• channel notes if demand differs by channel

Simple spreadsheet structure

A basic SKU replenishment planning sheet might look like this:

That final column is where the process needs to end: a practical order quantity by SKU.

Data hygiene checklist

Before you trust any forecast, check:

• Are all variants using one consistent SKU logic?
• Is sales history split correctly by SKU, not merged by family?
• Are returns, replacements, and internal transfers excluded or tagged?
• Is stock on hand sellable stock, not total physical stock including quarantine or damaged units?
• Are inbound POs current and accurate?
• Are lead times based on current supplier reality, not old assumptions?
• Are MOQ and pack multiples confirmed with the supplier?
• Are discontinued or clearance SKUs removed from standard reorder planning?

Step 2: Separate true repeat demand from one-off sales and promo spikes

One of the biggest forecasting mistakes is treating every sale as normal demand.

That inflates the forecast and pushes the team into overbuying inventory.

What should usually be adjusted or excluded

Review sales history and flag:

• launch sell-in spikes
• deep discount periods
• bundle activity that pulled demand forward
• wholesale loading orders that are not recurring
• stockout recovery periods
• influencer or PR spikes
• seasonal peaks that should not be annualised

Example: promo sales should not drive the baseline

Suppose a SKU usually sells 25 units per week. During a two-week promotion, it sells 90 units per week.

If you average the last 8 weeks blindly:

• 6 normal weeks x 25 = 150
• 2 promo weeks x 90 = 180
• total = 330 units
• average = 41.25 units per week

That gives you a baseline more than 65% higher than normal repeat demand.

For repeat order forecasting, a better approach is to:

• exclude the promo weeks entirely, or
• normalise them back toward expected baseline demand

In this case, the operational repeat forecast is probably still closer to 25 to 30 units per week, not 41.

Useful demand views to compare

For each SKU, compare:

• last 8 weeks average
• last 13 weeks average
• last 26 weeks average
• average excluding promo periods
• average excluding stockout weeks

You do not need advanced software to do this. A disciplined spreadsheet often gets most brands much further than a single broad monthly sales number.

Step 3: Group SKUs by repeat behaviour, not just by product family

Many brands group by collection, product type, or season. That helps reporting, but it is not enough for SKU-level demand forecasting.

Variants inside the same product family often behave very differently.

Practical demand groupings

A simple, useful grouping is:

• Fast-repeat SKUs: frequent sell-through, reorder consistently, stockout risk is expensive
• Stable-repeat SKUs: predictable sales, moderate reorder cadence
• Slow-repeat SKUs: do sell again, but slowly enough that MOQ risk matters
• Irregular sellers: demand is lumpy, event-driven, or difficult to replenish economically

Example: 12 variants, but only 3 carry the range

A brand sells one core product in 12 colours.

After 9 months of order history, the team sees:

• 3 colours generate 68% of repeat sales
• 4 colours sell steadily but at much lower volume
• 5 colours move mainly during launches, bundles, or seasonal campaigns

If the brand replenishes all 12 colours evenly, cash gets trapped in the long tail.

A better reorder planning by SKU approach would be:

• keep deeper stock on the top 3 colours
• reorder the middle 4 less often, possibly with lower safety stock
• let the bottom 5 run shallower, or move to seasonal or limited status

This is not just forecasting. It is assortment control.

Why product family averages are dangerous

If all 12 variants average 15 units per week on paper, the team may buy each variant as though it behaves the same. In reality:

• top variant may sell 40 units per week
• mid variant may sell 12 units per week
• tail variant may sell 3 units per week

That is how brands end up stocked out on winners while carrying cartons of weak variants.

Step 4: Translate forecast demand into reorder quantities

This is where forecasting becomes operational.

A simple reorder logic is:

Expected demand during lead time + safety stock - available stock - inbound stock = reorder need

Basic formula

1. Calculate average weekly sales by SKU
2. Multiply by lead time in weeks
3. Add safety stock
4. Subtract current sellable stock
5. Subtract confirmed inbound stock
6. Round for MOQ and pack size rules

Example: 6-week lead time plus 2-week safety stock

A core SKU sells 50 units per week.

• average weekly sales = 50
• lead time = 6 weeks
• safety stock = 2 weeks
• current stock = 180
• inbound = 0

Demand during lead time:

• 50 x 6 = 300 units

Safety stock:

• 50 x 2 = 100 units

Target coverage:

• 300 + 100 = 400 units

Reorder need:

• 400 - 180 = 220 units

If pack size is 20 and there is no MOQ issue, recommended order becomes 220 units.

If pack size is 50, you round to 250.

This is the practical link between how to forecast repeat orders and what eventually goes onto the purchase order.

How MOQs, pack sizes, and case quantities distort the forecast

The forecast number is rarely the order number.

Supplier constraints often force a larger purchase than demand alone would suggest.

Common constraints that change the final buy

• MOQ per SKU
• MOQ across a product group or PO
• pack size multiple
• inner carton quantity
• master carton quantity
• minimum production run by colour, size, or artwork version

Example: forecast says 420, supplier says 500

A SKU forecast shows reorder need of 420 units.

But the supplier requires:

• MOQ = 500 units
• pack size = 100 units

You cannot order 420. You must order at least 500.

That extra 80 units may be acceptable for a core SKU. It may be dangerous for a slow mover.

This is why MOQ-aware reorder planning should not be separated from demand review.

Operational decision rules

When MOQ exceeds forecast need, ask:

• Is this a core SKU with reliable repeat demand?
• How many weeks or months of stock will the MOQ create?
• Will the overage consume cash needed for faster movers?
• Can the supplier combine variants to meet a group MOQ?
• Should the SKU move to lower stocking, seasonal, or make-to-order status?

A large MOQ does not automatically mean “buy anyway.” Sometimes it means “do not replenish yet.”

How many SKUs should you actually keep in range?

Forecasting should not just tell you what to buy. It should help you decide what deserves ongoing stock commitment.

As SKU counts grow, range complexity often grows faster than repeat demand.

Signs a SKU may not deserve standard replenishment

• repeat sales are intermittent or weak
• demand only appears during discounts
• MOQ creates too much excess stock
• the SKU repeatedly ties up warehouse space
• forecast error is consistently high
• another variant serves the same customer need more efficiently

Fast-moving core SKU vs slow seasonal SKU

Compare two examples:

The same MOQ behaves very differently depending on sell-through.

For the core SKU, 300 units may be normal coverage.

For the seasonal SKU, 300 units may create many months of stock and unnecessary cash exposure.

A simple repeat-order planning workflow for brands

Most teams do not need heavy forecasting software to improve. They need a repeatable review process.

Weekly or monthly workflow

1. Export SKU-level sales, stock on hand, and inbound POs
2. Exclude or tag promo, launch, and unusual spike periods
3. Update average weekly sales by SKU
4. Classify SKUs by repeat behaviour
5. Calculate demand during lead time
6. Add safety stock by SKU group
7. Subtract available and inbound stock
8. Apply MOQ and pack-size rules
9. Review exception SKUs where MOQ creates excess cover
10. Approve order quantities and convert them into supplier POs

When you are ready to issue orders, the handoff should be clean. Use a structured document with correct units, pack details, and variant quantities. These guides help connect forecast output to execution:

• How to Write a Supplier-Ready Purchase Order
• Purchase Order Basics

Roles in the process

A practical division of responsibility often looks like this:

• Sales / commercial: flags planned promotions, launches, and channel changes
• Operations / supply chain: owns the forecast logic and stock review
• Purchasing / sourcing: confirms MOQ, lead time, and supplier constraints
• Warehouse / 3PL coordination: validates case, carton, and inbound handling assumptions

If 3PL or warehouse complexity is growing along with SKU count, How to Prepare Product Data for 3PL Onboarding is useful groundwork.

Common forecasting mistakes that lead to overbuying

The failure points are usually simple, not mathematical.

Common mistakes

• averaging across the whole product family instead of forecasting each variant separately
• using revenue instead of unit demand
• including launch spikes as normal repeat demand
• ignoring stockout periods that suppressed sales history
• combining DTC, wholesale, and promo demand without context
• buying to MOQ without checking weeks of cover created
• applying the same safety stock rule to every SKU
• replenishing weak variants because the range looks balanced on paper
• using outdated lead times
• keeping poor master data, so the same SKU is represented inconsistently across systems

A repeat-order forecast should reduce complexity, not hide it inside one average.

How better product data improves forecasting accuracy

Forecasting quality is heavily limited by product data quality.

If the team cannot trust SKU identity, pack hierarchy, supplier assignment, or lead-time fields, even a sensible forecasting model breaks down.

Master data fields that matter most

• SKU code and variant attributes
• unit of measure
• inner pack, case pack, and carton quantity
• supplier by SKU
• standard lead time
• MOQ by SKU or product group
• reorder point or planning threshold
• active, discontinued, or seasonal status
• barcode and packaging version where relevant

This is less about analytics sophistication and more about operational discipline.

One reason brands centralise this data is so planning, purchasing, and logistics teams are all working from the same version of the truth.

If your current process lives across several spreadsheets, inboxes, and supplier chats, consolidating SKU, pack, supplier, and reorder fields can make forecasting much more consistent. That is the kind of operational problem SKUWorks is built to support, but the principle applies even if you are still running the workflow manually.

FAQ

How do you forecast repeat orders by SKU?

Use SKU-level sales history to estimate normal repeat demand, exclude one-off spikes, calculate expected demand during lead time, add safety stock, then subtract current and inbound stock. After that, adjust the result for MOQ and pack multiples.

What data do I need to forecast reorder quantities by SKU?

At minimum:

• SKU-level unit sales history
• current sellable stock
• inbound stock
• lead time
• safety stock target
• MOQ
• pack size or case multiple
• active or discontinued status

How do MOQs affect SKU-level forecasting?

MOQs do not change demand, but they do change the final purchase quantity. A SKU may only need 420 units based on forecast demand, but if MOQ is 500 and pack size is 100, the order becomes 500 units.

What is the difference between reorder point and forecast demand?

Forecast demand is the expected future sales for a SKU. Reorder point is the stock threshold that tells you when to place an order, usually based on demand during lead time plus safety stock.

How often should product brands update their repeat-order forecast?

Most brands should review it weekly or monthly, depending on sales velocity, lead times, and supplier order frequency. Faster-moving ranges usually need more frequent review.

How do you forecast demand when you have many SKUs but limited sales history?

Start with short-term sell-through, classify SKUs conservatively, and avoid overcommitting on slow or unproven variants. Newer SKUs should usually carry lower confidence and lower stock commitments until repeat behaviour is clearer.

Should you forecast each variant separately or by product family?

Forecast each variant separately wherever possible. Product-family averages often hide the fact that a few variants drive most repeat demand while others sell intermittently.

How do pack sizes change the final purchase order quantity?

Pack sizes force rounding. If reorder need is 220 units and the supplier only ships in packs of 50, you must order 250. This is why pack-size logic matters in real replenishment planning.

Conclusion: use repeat-order forecasting to reduce stockouts and dead stock

The practical goal of this work is simple: buy the right SKUs, in the right quantities, at the right time.

That means moving beyond total sales and looking at:

• repeat demand by variant
• demand during lead time
• safety stock by SKU behaviour
• MOQ and pack constraints
• whether every SKU still deserves range space

For product brands, this is one of the clearest ways to improve cash flow, reduce avoidable stockouts, and stop filling the warehouse with inventory that only looked justified in a blended average.

You do not need a perfect forecasting model to get value. You need clean SKU data, a sensible review cycle, and a consistent way to turn demand into supplier-ready order quantities.