Cash

A slow SKU and a dying SKU look identical on the reorder list

A reorder list is a list of things you have decided to keep selling, and at the bottom of it that decision gets made by default. Four numbers you already own separate a SKU that is merely slow from one that is dying: the trend in units per week, the cover in your stockroom, what that stock costs to keep, and what the last order cost to land. The two SKUs below are invented worked examples.

A reorder report’s quantities are rarely controversial at the top of the list. The bottom is different. Down there the quantities are small, the rows look harmless, and each one is a decision to spend money keeping a product alive — a decision nobody makes out loud. The row appears, the quantity is modest, it is approved with the rest.

Step 1 — Separate the rate from the direction

One exclusion before the arithmetic: a SKU with a known seasonal pattern is a different question and is out of scope for this post, and neither SKU below has one.

Two SKUs, 24 weeks of sales each, in six four-week blocks so one week cannot carry the story:

The same history read three ways:

Both sell 4.0 a week now. One arrived there from 10.67 by shedding half its rate in six months; the other has been at 4.0 throughout. A long average cannot see that: over 24 weeks SKU A reads as an 8.17 a week product, because it still carries the 48 and the 44 from six months ago.

On a falling SKU, ordering from the long average is not the cautious choice. It is the aggressive one, wearing the clothes of the cautious one.

Three guards on the ratio. It compares equal windows, so shorten one and you shorten both. A SKU selling two a month throws ratios that swing on a single order. And a falling ratio tells you that demand changed; it does not tell you why — if seasonality, a promotion, a stockout or a channel change could explain the drop, resolve that before treating it as decline.

Step 2 — Turn the rate into how long your stock would take to clear

Both SKUs hold 260 units, off the same reorder list on the same day. Cover is on-hand units divided by units per week, so each rate above gives a different answer:

SKU B’s figures agree, because nothing about SKU B is moving. SKU A’s differ by a factor of two, and the friendlier one is what a long-average rule will use. Sixty-five weeks is fifteen months, long enough for the product to stop selling while you wait for the number to come down — and it is the optimistic reading, because it assumes the decline stopped today.

Cover looks forward; sell-through rate looks back at a finished period.

Step 3 — Put a price on the wait

Cut-offs get copied out of articles because the cost of waiting feels invisible. It is on invoices you already receive. SKU A landed at $9.40 a unit, so 260 units on hand are $2,444 of cash. A year of keeping them:

That is $526 ÷ 260 = $2.02 per unit per year, or 21.5% of landed cost a year — outputs of this example, not benchmarks. The third line is your judgement: use the rate that is economically relevant to your business — the rate you pay to borrow if this stock is financed on a line, or the return that cash would otherwise have earned if it is not. Pick higher and every wait gets more expensive and the case for clearing strengthens; pick lower and the stock has earned more patience.

Two costs are missing on purpose. Shrink and damage belong in your version if your count adjustments can measure them. And if the stock sits in a marketplace’s warehouse, that marketplace charges storage and may charge more for stock held a long time; look those up against your own account, because a copied figure is worse than none.

Apply it. At a flat 4.0 a week the 260 units run to nothing over 65 weeks, or 1.25 years, averaging 130 units: $2.02 × 130 × 1.25 = about $328. If pallet storage remains fixed as units decline, the true figure will be higher than this smooth-average estimate.

What the last order cost to land, and what it does not justify

The $9.40 is landed cost: supplier price plus that unit’s share of freight, duty and prep, worked out the way landed cost per unit sets out. It is the denominator above, and it is the starting point for what a repeat order would cost: the reorder decision should use the current expected landed cost, because if freight has moved, the next 260 units are not a $2,444 decision.

What it must not do is set a floor under the price you accept. That $2,444 left your account months ago. A unit that landed at $9.40 and will not sell above $8 is worth $8 today; the live question is whether $8 now beats what it fetches after another year on the shelf, net of what that year costs.

Step 4 — Four decisions, and the arithmetic that picks between them

Slow movers get treated as one problem with one answer. They are four, and Steps 1 to 3 pick between them.

Decision 1 — Reorder less

SKU A’s reorder quantity depends entirely on which rate feeds it. Over a 12-week coverage window:

Fifty units and $470 of new cash into a SKU already fifteen months long, produced by the window the average was taken over. Reorder less means using a recent rate that reflects the change rather than blindly carrying the long average forward.

Decision 2 — Reorder later

Order nothing this cycle, and write down when you will look again. Neither SKU is due at 65 weeks of cover. The useful date is where cover falls to your lead time plus your order cycle plus whatever safety stock you keep — the calculation behind which products to reorder first, read as a date instead of a rank. A zero somebody chose looks nothing like one nobody noticed.

Decision 3 — Discount to clear, at a discount you derived

Here is where published advice hands you a number and where you should refuse it. SKU A sells at $24.99, and $5.39 goes on costs that happen only because the unit sold — channel fees, pick and pack, the outbound label. That leaves $19.60 reaching the account. Part of the $5.39 moves with the price, so a discounted unit is slightly cheaper to sell than this allows.

The $2,444 already spent is in neither branch, because it is spent in both:

The second branch assumes 80 units are still on the shelf after two years, so stock averages 170 and holding costs $2.02 × 170 × 2. Clearing sells all 260 now at price P: cash in is 260 × (P − $5.39), with nothing left to hold. Solve for P against each branch:

Same SKU, same 260 units, same week. The whole distance between a 5% markdown and a 35% one is one judgement: how much of the tail is real. A fixed days-of-supply rule makes that judgement for you, knowing nothing about your holding cost. The arithmetic will not make it either — it tells you what your answer is worth, in dollars.

At $16.32 the unit still generates $10.93 of incremental cash after the $5.39 of selling costs; the $9.40 landed cost is sunk for this decision. The two rows are also a sensitivity test: if you cannot choose between 0 and 80 units of tail, work both ends.

Three caveats, because a break-even price is not a plan:

• A discount is an offer, not a clearance. Nothing here guarantees 260 units move at $16.32.
• A deep cut can move what buyers expect to pay for this SKU and the ones beside it. If you intend to keep selling it, that cost belongs in the branch too.
• Clearing in tranches buys evidence and costs time, and Step 3 priced the time.

Decision 4 — Stop

Stopping becomes the economic answer when continued replenishment no longer beats the alternatives; it can also be a strategic assortment decision — shelf space, a supplier minimum, the support the product costs, or a simpler range. Three things have to be written down:

1. What happens to the units on hand. Liquidate, bundle, donate, dispose. Each has a different cash figure and date, and the holding meter runs until one happens.
2. The reorder line itself. A SKU ended as a product but left on the reorder report comes back, because the report is a rule and the decision was a conversation. Record the stop where the buying happens, not only where the listing lives.
3. A date to read it back. If you discounted first, did the rate move?

SKU B does not show the same evidence of decline, so discounting is not justified by trend alone. Its ratio is 102%, its cover figures agree, and its problem is the order size: 260 units at 4.0 a week was a 65-week buy the day it arrived.

In SKU Compass, the two rates this method runs on are available as Supply Analysis report columns — Sales/Day and Days of Supply — and the toolbar carries a Hide no-sales SKUs checkbox, which for this exercise is the filter to leave switched off.

Where this leaves the reorder list

The series opened on which products to reorder first, which is a sort: everything on the list is getting bought and the only question is the order. This is the same list read from the other end, where the question is whether the row belongs on it.

The trend says whether the recent weeks are weaker than the weeks before them. The cover says how long you are committed for. The holding cost says what that commitment costs a year, in money on an invoice. The landed cost is where the estimate of the next order starts. Run the four and the row turns back into a decision: buy less, buy nothing this cycle, take the cash, or end it.