Every brand that reaches a shelf eventually leaves one. The question is never whether, it is which of three exits you are in — because they have different dates, different costs and a different person on the other side of the conversation.
This is the third page of a sequence. Retail readiness covers the first delivery, the second order covers what decides a reorder. Both assume the product stays. This one does not: it works through what an exit costs, which route recovers the most, and why in this category the stock usually cannot come home at all.
Three exits, and only one of them is announced
The first is a delisting: the retailer removes the SKU from the next range at a category review and tells you on a calendar you did not set. The product still exists, still sells somewhere, and may come back. The second is a discontinuation: you end the SKU yourself — end of life, a mould you will not run again, a component that is no longer available. Because you control the date, this one is usually the cheapest of the three. The third is the one nobody names, and the most common: a listing that never became a business. It was set up, received, never reordered, and the exit is silent rather than announced.
All three end with the same two questions. What happens to the units, and which route recovers the most of what they cost — not what they were worth on a shelf, what they cost landed. The difference between those two numbers is the entire subject of this page.
One thing is worth fixing before the rest of it. An exit is a date, not a mood. A brand that says "it did not work out" has learned nothing it can act on; a brand that says "the shelf pull is 14 November, 360 units are unsold, and the markdown share is capped at 12% by the vendor agreement" is running a decision with numbers in it. Everything below is aimed at the second sentence.
The notice is a calendar, not a verdict
A delisting notice is short and most of it is dates. It names the SKU, the reason code the retailer's system uses, and the last day it will accept a purchase order; everything after that is a countdown. Read it as a schedule, because six of the dates in it are operational and only two are ever negotiable.
A delisting notice is a calendar before it is a verdict: almost everything in it is dates, and dates can be prepared for.
The shelf exit, counted from the notice
Six dates from the notice, and only the first two are negotiable. The shelf pull is set by the planogram, not by your stock position.
The negotiable two are the final order cut-off and the receipt window, and they are negotiable for a commercial reason rather than a sentimental one: until the reset, the retailer has shelf space committed to your product, and an empty bay costs it more than a late pallet. Everything after the reset is arithmetic. The shelf-pull date is set by the planogram, not by your stock position; the return window is set by the vendor agreement; and the last possible deduction date is set by the payment terms — which is why a delisting can still cost you money a hundred days after the last unit left the shelf.
Two habits pay for themselves here. Ask for the notice in writing with the reset date on it, because a verbal "we are probably taking it out in spring" cannot be planned against and cannot be shown to anyone. And put the exit terms into the vendor agreement at setup rather than negotiating them at exit: who pays return freight, what condition stock must be in to come back, whether unsold units are returned, marked down or destroyed, and what share of a markdown the brand carries. That clause is worth more than a point of unit price, and it is the one nobody reads while the listing is going well.
Why this category usually cannot take its stock back
In most categories a delisted line comes home. The retailer ships the unsold cartons back, you credit it, and the goods go to an outlet, a jobber or a charity. In this category that route is largely closed, and the reason is the product rather than the contract.
A device that has been on a shelf cannot be certified as untouched, and the return policies written across this category rest on exactly that distinction: sealed and unopened only, inside a narrow window, and nothing at all once the seal is broken. The same logic runs upstream. A brand cannot restock a carton it cannot certify either, so the stock that does not sell usually stays where it is and is marked down in place — which converts an inventory problem into a commercial negotiation about who funds the discount, and that is the subject of the two sections after this one.
There is a second consequence, and it changes how a launch should be sized. Consumer returns in this category are destroyed rather than restocked, so the true cost of a return is the full landed cost of the device and not the margin on it. That is the honest argument for a smaller first run: a small run is cheap to be wrong about, and being wrong is a normal outcome of a first listing rather than a failure of the brand.
The arithmetic of one more month
Holding dead stock feels free because nothing is invoiced. It is not free; it is prepaid. Inventory carrying cost is the sum of capital, storage, insurance, handling, obsolescence and the risk of damage, normally expressed as an annual percentage of inventory value. The number usually quoted is about a quarter of inventory value a year, and its traceable origin is a 1995 trade-press estimate that put the range between 25% and 55% depending on the components counted [Richardson, 1995]; logistics textbooks have used 25% as a first approximation ever since. It is an estimate rather than a measurement of your business, and it is still the right shape for the decision.
Parked stock invoices nothing, which is why it looks free. What it does is charge in advance.
The cost of not deciding, on a thousand parked units
Twelve months of waiting costs $1,600 — exactly what ten points of wholesale price would cost across the whole run.
Applied to one run, the arithmetic is blunt. A thousand units at $6.40 landed is $6,400 of inventory; at 25% that is $1,600 a year, or about $133 a month, whether or not a single unit moves. Twelve months of waiting therefore costs the same as giving up ten points of the wholesale price across the entire run — $1.60 a unit, $1,600 on a thousand. That equivalence is the argument, and it is not a rhetorical one: the deeper markdown converts stock into cash and closes the file, while the year of waiting converts stock into another year of waiting and leaves the same units on the same shelf.
The test that decides it is not the size of the discount, it is whether the constraint was ever the product. If the answer at the review was a lost facing, a price point that did not fit the fixture, or an item that never made it out of the back room, more time changes nothing — the problem was distribution, not velocity. Holding is only rational when the reason for slow movement has an end date: a season, a reset, an approval that is genuinely coming.
Markdown money: who pays for the discount
When the retailer clears your SKU it takes a discount on the shelf price and, under most vendor agreements, sends you a share of it. The name for that share is markdown money, or a markdown allowance, and it behaves like a chargeback: it is deducted from your invoice rather than invoiced to you, which is why it is easy to miss until the remittance advice arrives short. The retail-management literature treats it as a normal cost of a retail relationship rather than as an exception [Levy & Weitz, 2014].
When the retailer clears the shelf it does not send the goods back — it sends you a share of the discount. That share is the number this figure plots.
Your share of the markdown, on 360 unsold units
At a 40% markdown your share is $2,304 — exactly the landed cost of the 360 units being marked down.
The arithmetic is worth doing before the conversation instead of after it. On a $32.00 shelf price, a 40% markdown is $12.80 a unit, and across 360 unsold units a 50% share is $2,304 — which is exactly the landed cost of those same 360 units at $6.40. At that point the brand has funded the discount with the entire value of the goods the discount was applied to, and at 55% it is more than that.
Three things cap the exposure, and all three belong in the agreement rather than in the argument. A maximum rate per unit or per season, stated as a number rather than as a principle. A requirement that a markdown be notified before it is taken, so that it is a decision and not a discovery. And a definition of which stock it applies to — the units you shipped into that listing, not the units the retailer redistributed to three other stores. A brand that negotiates two points off the unit price and accepts uncapped markdown support has bought a spreadsheet rather than a margin, which is the same point the deduction section makes about chargebacks: both are percentage deductions on an invoice that grows.
One asymmetry decides the tone of the meeting. The retailer has already decided, and the markdown is in its system before anyone calls you. Your leverage is therefore not whether it happens but what the number is, and the strongest thing you can bring is your own sell-through calculation from the weeks of cover on the same stock being marked down. A brand that arrives with arithmetic is negotiating; one that arrives with a story is being informed.
There is a documented case for auditing the line rather than trusting it. In 2005 Saks disclosed that one of its merchandising divisions had collected markdown allowances it was not owed, and its own filings put the figure at $26 million for fiscal 1999 to 2003, with a further $8.2 million for the three years before that; the money was repaid. The mechanics matter more than the scandal: the vendor had shipped the goods, the discount was applied in a system the vendor could not see, and the deduction arrived on a remittance advice. Photographing the bay and reconciling the remittance, rows five and six on the sheet further down, are what make the same thing visible at a thousand-unit scale.
If the stock does come back, what each route pays
When the units are in your warehouse rather than on a shelf — a return that was accepted, a direct-to-consumer brand ending a line, a distributor closing an account — the question becomes which route recovers the most, and the routes are not close together.
If the goods do come back to your warehouse, these are the five doors and what each one pays per unit.
What each exit route pays, per unit
From $19.20 to minus $0.35 a unit: all five exits fit inside twenty dollars, and only one of them leaves cash.
Read the ladder from the top down and the shape is the same in every category: the route closest to the end customer pays the most and takes the longest, and every step away from that customer costs margin. Selling the remaining stock at 60% of the shelf price through your own channel recovers $19.20 a unit but needs traffic a device that already failed to sell does not have. An off-price partner at 40% pays $12.80 and takes the stock in one delivery. A jobber bid at $6.40 is your landed cost, which reads like breaking even and is actually the loss of the entire gross margin plus the freight you already paid to bring the goods in.
Order matters more than the individual rates, and the reason is that a clearance reprices the market before the stock arrives. Once the retailer has run your device at 40% off, a resale buyer who was quoting 40% of the shelf price is now quoting against a market where the shelf price is $19.20, and the next bid reflects it. The practical rule is to price the secondary route before the clearance and to run both against the same calendar, because a bid accepted after a markdown is a bid against a different market.
Two routes at the bottom of the ladder deserve plain language rather than optimism. Donation recovers nothing in cash, may recover something in tax depending on jurisdiction and structure, and is often unavailable here at all, because a device that cannot be resold to a customer usually cannot be given to one either. Destruction is the honest end of the ladder and costs money per unit, which is the argument for taking the decision while the stock still has a door rather than after it has none.
What survives the exit, and what is retired with it
An exit destroys an SKU, not a business, and it is worth separating the two lists — because in one direction the list is shorter than brands expect and in the other it is longer.
What survives: the mould, if it is yours, which is the argument in the OEM tooling section for paying for a tool outright rather than amortising it into a unit price. The construction knowledge, which is what makes the next SKU cheaper to develop than this one was. The factory relationship, if the exit was commercial rather than a quality failure. And the measured data, if the device ever went through a bench — the only asset on the list that does not decay, and the one a buyer will ask for again.
What does not survive: the GTIN, which is retired rather than reassigned, because a barcode that once identified a dead product becomes a permanent ambiguity in every retailer item file and every marketplace catalogue. The artwork and the packaging plates, which are tied to a pack size designed for a planogram that no longer exists. And the certification file, which is tied to a model and a bill of materials — change the motor supplier and the report describes a device you are no longer building. That last point is the one that catches brands re-entering with an old test report in hand: it is not evidence about the new build, and a retailer's compliance team knows it.
The five questions, while the data is still complete
The exit is the only moment when the data is whole: you know the sell-through, the facings, the price point, the markdown and the deduction total, and nobody is defending anything yet. It is also the moment when everyone wants to move on. Run the post-mortem anyway, on one page, with five questions.
| Question | Where the answer lives | What it changes next time |
|---|---|---|
| Did it reach the shelf at all? | The planogram and the first four weeks of receipt data | If it did not, the problem was distribution — a new product will not fix that |
| What was velocity against the category, not against the forecast? | The retailer's sell-through report, beside your own plan | Tells you whether the product or the number was wrong |
| Did the price point fit the fixture? | Shelf price against the band the bay already carries | Sets the price test for the next launch, before the packaging is drawn |
| Which criterion lost the review, in the buyer's words? | Their answer, in writing, asked for on the day | Turns an opinion into a specification you can design against |
| What did the exit cost, all in? | Markdown share, freight, destruction, unrecovered tooling | Sizes the next first order, because the exit is never in the launch budget |
The last row is the one that changes behaviour, because the exit cost is almost never in the launch budget. A first order at $6,400 landed that ends with a $2,304 markdown share and $1,900 of unrecovered tooling was a $10,600 experiment — and a brand that knew that before the launch would have sized it differently. That is the same argument the launch cost figure makes from the other end, before the money is spent rather than after.
The exit sheet
Seven rows in the order the money leaks. The cheap rows are at the top and the expensive ones are at the bottom, which is why the two that belong before the first purchase order are the two nobody does.
The shelf exit, from notice to post-mortem · SirenPicks · sirenpicks.com/exit/
This sheet is built to print as one page: Ctrl-P (⌘-P) drops the navigation, the figures and every other section, and leaves the seven rows on a single sheet you can take into the exit meeting.
| When | What has to be done | What it protects |
|---|---|---|
| At listing, before the first purchase order | Negotiate the exit clause: return basis and freight, a markdown cap, notification before a markdown, and which units it covers | The four numbers that are no longer negotiable at exit |
| Week 1 of the notice | Put the six dates on a calendar, confirm the reset date in writing, and reply formally | Turns a verdict into a schedule you can plan a last order against |
| Before the final order cut-off | Size the last order from cover weeks, not from the hope of a turnaround | Avoids a final run sized by the calendar instead of by demand |
| Two weeks before the shelf pull | Request the markdown in writing with units and depth, and check it against the cap | Converts a future deduction into an agreed number |
| At the shelf pull | Photograph the bay and count the units that came off it | The only evidence that the markdown applied to the stock you shipped |
| Days 0–30 after the last receipt | Reconcile the remittance advice against the purchase order, line by line | A deduction disputed inside the window is a claim; one found later is a write-off |
| Within 60 days | Run the five-question post-mortem and decide which SKU inherits the tooling | Stops the same tuition fee being paid twice on the next launch |
Figures in the sections above are arithmetic on assumptions printed in their captions. Carrying rate after Richardson (1995); no retailer rate card is quoted anywhere on this page.
A last note on the sheet itself: the first two rows are worth more than the other five combined, and they are the two that have to be done when nothing appears to be wrong. An exit clause negotiated during a good quarter costs a conversation; the same clause negotiated during a delisting costs whatever the retailer decides it costs.
What we can and cannot tell you
We can show the arithmetic of a markdown, a carrying cost, a recovery ladder and a deduction, and we can read a pack drawing against a planogram template. We publish every field the factory declares in the specification index, and the measured side of the bench in the reviews. What we cannot do is see your vendor agreement, know your cost of capital, or tell you what a specific retailer will accept in a specific negotiation: those belong to your account, and a page that guessed them would be inventing the most consequential numbers in the sentence.
If the question is still the first delivery, start with retail readiness. If a reorder is still on the table, the second order is the page before this one. If the question is what the exit commits you to on the manufacturing side, OEM and private label sets out what each route costs to change, and the startup guide covers the sequence that comes before any of it.
This page is published by the SirenPicks independent review desk, which tests devices and takes no payment for a placement, a ranking or a score. No retailer's rate card, threshold or markdown policy is quoted here: the mechanics are documented and the numbers are yours to substitute.
Sources cited
- Richardson, H., “Control Your Costs Then Cut Them”, Transportation and Distribution, December 1995 — the inventory carrying cost estimate of 25–55% of inventory value a year, with the component breakdown, that later literature still reproduces.
- Chopra, S. & Meindl, P., Supply Chain Management: Strategy, Planning, and Operation, Pearson, 2016 — cycle stock, safety stock and the cost of holding inventory as a decision rather than a residual.
- Silver, E. A., Pyke, D. F. & Thomas, D. J., Inventory and Production Management in Supply Chains, CRC Press, 2016 — obsolescence and the treatment of stock whose demand has stopped.
- Levy, M. & Weitz, B. A., Retailing Management, McGraw-Hill, 2014 — markdown management, markdown money and the merchandising measures a category review uses.
- Berman, B. & Evans, J. R., Retail Management: A Strategic Approach, Pearson, 2013 — shelf space as the resource a buyer controls, and the mechanics of a delisting.
- Guide, V. D. R. & Van Wassenhove, L. N., “The Reverse Supply Chain”, Harvard Business Review, February 2002 — why the value recovered from returned goods falls as the product moves further from the end customer.
- Saks Incorporated, disclosure of improperly collected vendor markdown allowances, 2005 — $26 million for fiscal 1999–2003 and $8.2 million for 1996–1998, per the company’s SEC filings as summarised in SPS Commerce’s deduction-audit guide (2026). Cited only for the mechanism: an allowance collected on stock the vendor had already shipped.
- Wikipedia, 《Planogram》 — the shelf drawing that decides the pull date, and the facings a review argues about.
No industry average appears on this page. Every figure is arithmetic on assumptions printed in its own caption, and the one published estimate — the carrying rate — is attributed rather than presented as a measurement of your business.