Answer in brief
A stockout on a marketplace switches off more than sales: it switches off the position the listing spent months earning. Here is how lead time, storage cost and the season window combine into a restock plan that does not strand cash in a warehouse.
What you actually lose when the stock runs out
The direct answer first: the cost of a stockout is not the revenue you missed on the empty days. It is the search position the listing spent months earning, and buying that position back is usually more expensive than the extra case of goods would ever have been.
Position on a marketplace is a cumulative result built from impressions, clicks, orders and reviews. An item nobody can buy is a poor use of a search slot, so the platform has every reason to show an available competitor instead. Sales stop before the last unit ships.
When the restock finally lands, the counter does not resume where it stopped. The listing re-enters search with a weaker recent signal and competes against sellers who never went dark. Recovery is measured in weeks, while going to zero took a matter of hours.
That changes the planning question. You are not holding stock that is merely enough to cover orders, you are holding enough that the listing never switches off, and the comparison is between the cost of a few surplus units and the cost of rebuilding position.
Why an empty listing loses position, not only orders
A marketplace sells shopper attention. Showing a product nobody can buy wastes a slot and spoils the visit, so the platform's incentive matches any shopkeeper's: the empty shelf gets pulled and something sellable takes the space it was occupying.
Then the feedback loop takes over. Fewer impressions mean fewer clicks and fewer orders, which weakens exactly the signals a ranking is built on. Zero inventory starts a slide that keeps running for a while after the goods are physically back.
External sources suffer separately. Links, roundups and ordinary search engines keep sending people to a page where nothing can be bought. A share of that traffic never returns, and the page collects a high exit rate that is awkward to unwind later.
So the honest comparison is not lost revenue against storage cost. It is lost revenue plus the cost of recovering position, weighed against storage cost. That second version of the sum changes most restocking decisions all on its own.
Lead time is the planning unit, not a date in the calendar
Reordering when stock gets low is the most common failure, because low is not a quantity, it is a duration. The only useful question sounds different: how many days pass between pressing order and having a unit that a customer can actually buy.
That window holds far more than production and shipping. Add approval, payment, packing to the platform's requirements, labelling, the warehouse receiving window and the time until the goods are genuinely listed as available. Each stage is an independent source of delay.
Plan against the worst realistic case rather than the average. An average lead time is a promise that gets broken at precisely the moment demand is above normal and a delay is costing you more than it would in any quieter week.
One practical habit pays for itself: record the real, measured lead time of every delivery you receive. After a few cycles you have your own numbers instead of a supplier's estimate, and the reorder point stops being a guess dressed as a decision.
Three clocks: supply, sales velocity, and the season window
Inventory is governed by three clocks that run independently of each other. The first is lead time. The second is sales velocity, the number of units leaving per day. The third is the season window, the period in which this demand exists at all.
Multiplying the first two gives the level you must not fall below: daily velocity times lead time, plus a buffer for variability. That is a reorder point, and it is a very different object from the feeling that about a third of a box is left.
The third clock decides whether to order at all. If the season window will close before the shipment can arrive and sell through, the disciplined answer is to place no order and let the remaining stock run down in a controlled, deliberate way.
Trouble starts when three different people read the three clocks. Purchasing watches lead time, sales watches velocity, and nobody owns the season, which is how money ends up sitting in a warehouse during the week after the peak has passed.
Storage is rent you pay on a wrong forecast
Storage inside a marketplace fulfilment network is paid for, and the rate usually climbs with both the time held and the volume occupied. That makes a surplus unit something other than free insurance: it is an asset that gets slightly worse every day it sits.
The rate is only part of the bill. Cash tied up in goods is not working anywhere else, products age or get scuffed in handling, and a seasonal item that outlives its window can often only be cleared at a discount. Three different losses of three different sizes.
This gives the problem its real shape. The goal is neither maximum stock nor minimum stock, but the level at which the expected cost of holding is roughly equal to the expected cost of running out. The optimum lives between the two instincts.
Cheap items with short, reliable lead times deserve a generous buffer. Expensive, bulky or sharply seasonal ones deserve a thin buffer and more frequent ordering. The same warehouse needs different policies for different lines in the same catalogue.
How a season distorts the arithmetic
A season is not just a peak. It is four regimes, the ramp, the peak, the decline and the tail, and both velocity and the cost of being wrong differ enormously between them. The reorder point has to be recalculated at every transition.
During the ramp a stockout does the most damage, because that is when the listing earns the position it will trade from during the peak. Going to zero in those weeks does not hand a competitor one week of sales; it hands them the whole season.
At the peak the risk inverts. The last purchase order often arrives after demand has already turned down, leaving you holding goods with no window left to sell them in. Closer to the peak, the sensible move is smaller batches ordered more often.
In the tail the objective changes again: leave the season with a stock level close to zero. Markdowns, bundles and switching off promotion all belong here, because storing seasonal goods for a year usually costs more than selling them cheaper today.
The hole in your data: never average through a stockout
Days with no stock appear in reports as days with no demand. Average across them and the forecast comes out low, so you order less, run out sooner, and the same error reproduces itself in the next cycle, a little more severely each time round.
The fix is unglamorous: exclude out-of-stock periods when you calculate sales velocity, and compute it only across the days when the item could genuinely be bought. The resulting number describes demand rather than describing your own logistics.
The same applies to sharp dips with other causes, such as a technical outage, a listing suspended for review, or a pricing mistake left live across a weekend. Flag those days explicitly, or they stay in your history forever as evidence of weak demand.
Keep a plain event log: the date, what happened, and how many days it lasted. It is a boring document that turns into the most valuable data you own about your own catalogue somewhere around its first birthday.
A restock plan measured in days of cover, not units
Order three hundred units means nothing until somebody answers covering how many days. Convert the target into days of cover: how many days of selling the remaining stock should support at the moment the next shipment is received into the warehouse.
A workable frame is that target cover equals lead time plus a buffer plus the interval between orders. The less reliable the supplier and the more expensive a stockout, the larger the buffer, and that buffer is a deliberate purchase rather than vague caution.
Then test the plan against money. Multiply the target level by unit cost and by the expected days held. If the number is unacceptable, do not shave the buffer first: shorten the lead time, or narrow the range of products you are trying to keep in stock.
Finally, write down when the number gets recalculated. A reorder point is not a constant. It moves when you change supplier, when storage rates change, and at every transition between the regimes of a season. Once a month is the minimum sensible rhythm.
What to do when restocking in time is no longer possible: A reorder point is daily sales velocity multiplied by…
When the shipment cannot arrive in time, the objective changes. You are no longer trying to sell the maximum; you are trying to protect the listing's position. Stretching the remaining units across more days is often worth more than clearing them in a weekend.
The first lever is price. A measured increase slows sales and extends availability without switching the listing off. The second is promotion: paying for traffic into stock that is about to disappear buys visits that you have nothing left to convert with.
The third is variant management. If sizes or colours sell down unevenly, it is usually better to restrict the exhausted variants and keep the listing itself buyable than to let the whole page fall to zero because of one unusually popular size.
The fourth is honest timing. A stated dispatch window you actually meet is far cheaper than cancellations and the complaints that follow them, and seller-side cancellations tend to damage account standing more than an openly declared delay ever does.
The listing on thin stock: availability, dates and honest labels
Availability is structured data, not just a line of text. In the Schema.org Product vocabulary, price and availability live on a nested Offer, whose availability property takes values such as InStock, OutOfStock, PreOrder or BackOrder. The markup has to agree with what the shopper sees.
A mismatch between markup and page is a genuine risk rather than a technicality: search engines can stop trusting your data, and the shopper arrives holding a promise the page cannot keep. Update the markup along with the stock status, not once a quarter.
The status message itself has to be accessible. WCAG 2.2 asks that meaning is not carried by colour alone, so only a few left should be written in words rather than signalled by a red dot, and dynamic messages need to reach assistive technology.
There is a content side too. Dispatched within three working days is more useful than shipping soon. A specific window reduces cancellations and chat questions, which indirectly protects the same ranking signals you are working so hard to defend.
Money frozen in a warehouse: stock problem or assortment problem
Not every unit gathering dust is a planning error. Sometimes the stock level was calculated correctly and the demand simply is not there. In that case the surplus is a symptom, and ordering a bigger batch only enlarges the loss you have already taken.
The test is straightforward. If an item sells steadily and hits zero between deliveries, that is a stock problem and a reorder point fixes it. If it sits still while fully available and receiving traffic, that is an offer problem wearing an inventory costume.
Offer problems are not solved in a warehouse. They live in price, bundle, photography, copy, category placement and the competitive set around the listing. A demand audit is useful here precisely because it separates lines worth stocking from lines worth retiring.
A useful discipline before any large purchase order: say out loud why the next batch will sell faster than the last one did. If the sentence will not finish, the money is better left in working capital than converted into pallets in a warehouse.
When to hire help, and what each package actually covers
Outside help is worth paying for once the question stops being arithmetic. You can calculate a reorder point yourself. The harder work is seeing where demand is being lost and deciding which lines deserve inventory in the first place.
The Demand Audit is $50 and takes 1-2 working days, includes one round of revisions on the findings document, and does not include access to your seller account. It is a reasonable step to take before committing to a large purchase order.
Listing Refresh is $70 across 2-3 working days with one round of revisions; photography, ad spend and account management are not included. The Marketplace + SEO System is $110, runs 3-5 working days and carries two rounds of revisions on listings and copy.
If you need a continuing rhythm rather than a project, Managed Demand Growth is $200/mo, works in a monthly cycle and stops on 30 days notice, with advertising spend on the marketplace excluded. Each package's terms are listed on /services/marketplaces.
Practical checklist
- Measure the real lead time from your recent deliveries instead of the supplier's estimate.
- Recalculate every line's reorder point in days of cover rather than in units.
- Exclude out-of-stock periods before you compute sales velocity.
- Cut batch size and order more frequently as the season approaches its peak.
- Check that the availability value in your Offer markup matches what the listing shows.
- State a specific dispatch window in words and confirm it reads without relying on colour.
Questions and answers
Which costs more: holding extra stock or running out?
Both sides need a number. Surplus units cost storage fees, tied-up cash and a possible markdown. Zero costs the missed orders plus the work of rebuilding position, and it is that second term that usually decides the question.
How much does the Demand Audit cost and what is included?
The Demand Audit is $50 and takes 1-2 working days. It includes one round of revisions on the findings document; access to your seller account is not part of the package.
How quickly does a listing recover its position after a stockout?
There is no fixed period, because it depends on the category, the competition and how long the listing was dark. Plan for weeks of active trading rather than days, and do not assume position returns the day the goods arrive.
How do I calculate sales velocity if the item was often unavailable?
Count only the days when the item could actually be bought, and log the out-of-stock periods separately. Otherwise the average drops, you order too little, and you repeat the stockout in the following cycle.
Should seasonal leftovers be marked down at the end of the window?
Compare the cost of holding the goods until the next season against the margin lost to a discount today. For bulky or fast-dating items the markdown usually wins; for compact, stable ones it often does not.