September is the only month in the second half of the year where you still have both information and time. By mid October the Q4 budget is largely spent, and by November you are managing what you already own. The first three weeks of the autumn season carry enough signal to direct the rest of the year, but only if you read them at the right level of detail. Most resellers read them at model level, which is the wrong level and produces the wrong decision.
The formula, and why the standard version is not enough
Sell-through rate is units sold divided by units received, over a defined period, expressed as a percentage. Sixty pairs in, thirty sold, 50% sell-through. Every inventory guide stops there, and that is where the useful part begins.
The problem is that footwear is not one product. A model in EU 39 to 46 is eight products that happen to share a name, and they do not share demand. Averaging them into one number destroys exactly the information you need to place a reorder. So the calculation you actually run is sell-through per size, per model, per week: units sold divided by units received, at size level.
If your platform will not export it, build it once in a spreadsheet with three columns, size, received, sold, and keep the format for every season. The setup cost is an hour. It is the highest return hour in the buying calendar.
The same 50%, two opposite decisions
Take a men's model bought 60 pairs deep across EU 39 to 46. Three weeks in, 30 pairs are gone. Here are two ways that can happen.
| Size | Received | Case A sold | Case B sold |
|---|---|---|---|
| 39 | 4 | 0 | 2 |
| 40 | 6 | 2 | 3 |
| 41 | 10 | 9 | 5 |
| 42 | 12 | 12 | 6 |
| 43 | 12 | 11 | 6 |
| 44 | 8 | 6 | 4 |
| 45 | 5 | 0 | 2 |
| 46 | 3 | 0 | 2 |
| Total | 60 | 30 | 30 |
Case A is a model with real demand and nothing left to sell. Sizes 41 to 43 are at or near 100%, and a customer who wears 42 will not buy 46. Every day this sits unreordered is demand walking away. You reorder immediately, in 41 to 44 only, and you do not touch the edges you are already holding.
Case B is a model selling evenly to a small number of people. Nothing is out of stock, nothing is urgent, and it will still be selling in December without a markdown. You leave it alone. A reorder here does nothing except extend the tail into January.
Same headline number. Opposite decisions. This is the whole argument for reading at size level.
Reading three weeks correctly
Three weeks is enough because the shape of the curve is established by then, not because the volume is meaningful. What you are looking for is direction.
- Week 1 is contaminated by launch effects, existing customer demand and your own promotion. Record it, do not weight it.
- Week 2 is the first clean read. Whatever sells in week 2 without support is real demand.
- Week 3 tells you the slope. Week 3 above week 2 means the model is building. Week 3 at 50% or less of week 2 means week 1 was your season.
Compare weeks against each other, not against a plan. The plan was written before you had data.
Weeks of cover, the second number you need
Sell-through tells you what happened. Weeks of cover tells you how long you have. Units remaining divided by average weekly units sold, calculated per size.
Case A above, size 42: 12 received, 12 sold, average 4 per week, 0 remaining. Weeks of cover: zero. That size has been dead for however many days have passed since it emptied, and no report flags it because it looks like a success.
Case A, size 46: 3 received, 0 sold, 3 remaining. Weeks of cover: infinite. That is not stock, it is a markdown waiting for a date.
The rule of thumb for the autumn window: a size under 3 weeks of cover is a reorder, a size over 12 weeks of cover is an exit. Anything between the two is working as intended and should be left alone.
The decision thresholds
Fixed thresholds beat judgment because they force the decision before you have grown attached to the stock. These are three-week numbers for a normal autumn footwear assortment and they should be adjusted to your own history after two seasons, not abandoned.
- Above 45% sell-through with the middle sizes cleared: reorder, depth in the middle of the run only, and reorder now rather than after week four. Lead time is the constraint, not availability.
- 25% to 45%, evenly distributed: no reorder. The model is working through its stock at the right pace and a second buy will simply extend the tail into January.
- 25% to 45%, concentrated in two or three sizes: reorder those sizes only, at low depth. This is a size problem, not a model problem.
- Below 25%: stop. No reorder at any depth. Plan the exit now, in November, at a controlled discount, rather than in February at whatever the market will pay.
- Below 15% after three weeks: the model is not underperforming, it is wrong for your customer base. Note the brand, the silhouette and the price point, and do not repeat the buy next season.
One rule that overrides all of the above: never reorder in the last two weeks before your peak dispatch window closes. Stock that arrives after the peak is January stock bought at Q4 prices. Courier capacity compresses before you expect it to, which is a separate planning problem and the reason winter buying decisions get made in summer, as covered in Winter Footwear Buying Starts in Summer.
The classic mistake
The error that costs the most money in September is reordering on the basis of what sold, rather than what is left.
It happens because sales reports are built around sales. You open the report, you see that a model moved 40 pairs, it is at the top of the list, and it feels like the obvious reorder. But the report does not tell you that those 40 pairs were the sizes everybody wears and what remains is 39, 46 and 47. Reordering the model as a block means buying a fresh full size run, which means buying the edge sizes again on top of the edge sizes you already hold. Two seasons of that and a large share of your capital is locked in sizes that only clear at a loss.
The inverse error is equally common: a model shows weak total sales and gets dropped, when in fact it sold out of its middle in week one and has spent two weeks unable to sell because there was nothing left to sell. That model looks like a failure in the report and is actually your strongest reorder candidate.
The fix is procedural, not analytical. Read the residual stock report first, then the sales report. Ask what is left, in which sizes, and how long it will take to clear at the current rate. Only then look at what sold. The order of the two reports changes the decision more often than any calculation in this article.
How the thresholds differ by category
The same numbers do not apply to every part of the assortment, because the selling window is not the same length.
- Core sneakers such as the Nike Air Force 1 Low, the Dunk Low, the adidas Samba or the New Balance 530 have a twelve month window. A 30% three-week sell-through is fine here, because nothing goes out of season. Reorder on weeks of cover, not on percentage.
- Trend silhouettes such as the Puma Speedcat, the ASICS GEL-Kayano 14 in seasonal colour or a collaboration release have a two to three month window. Apply the thresholds above strictly, and treat below 25% as terminal rather than slow.
- Seasonal boots and cold weather such as the UGG Tasman, Timberland 6-inch, Moon Boot and lined Crocs and Birkenstock variants have a six to eight week window and a hard end date. Here a 45% three-week read is not comfortable, it is the minimum to be on track, and the exit decision has to be made in November rather than deferred.
Running one threshold across all three is how retailers end up with a January warehouse full of boots and no core stock. The distinction between a seasonal buy and a permanent line is the same one that drives the assortment logic in Fall/Winter Footwear: How to Stock Smart and Sell Through the Season.
What to do with the exit list
The bottom two bands are not a failure to be hidden, they are a cash decision with a deadline. Three options, in order of how much they return.
- Bundle rather than discount. Pair a slow model with a fast one at a combined price. This protects the price positioning of the slow model, which matters if you intend to carry it again.
- Discount by size, not by model. If 41 to 43 sold out and 39 and 46 are sitting, mark down only the sizes that are stuck. Cutting the price of the whole model gives away margin on sizes that never needed it.
- Move it into a dropship or partner channel rather than holding it for a general sale. Getting a low turn reference in front of a different customer base beats another six weeks of shelf time. Q4 is the window where this actually works, and the timing argument is in Q4 Dropshipping: Why Retailers Must Prepare Now.
The September checklist
- Export units received and units sold by size, per model, for the last three weeks.
- Calculate sell-through per size. Mark every size at 80% or above as effectively out of stock.
- Calculate weeks of cover per size. Flag anything under 3 and anything over 12.
- Split models into the five threshold bands, adjusted for category window.
- Build the reorder list from the top two bands only, in middle sizes, and send it before the end of the month.
- Build the exit list from the bottom two bands and schedule the action for November, with the date fixed now.
- Leave the middle band alone. Doing nothing is a decision and in this band it is the correct one.
Q4 budget spent this way is spent on models that have already proven they sell to your customers, in the sizes those customers actually wear. Everything else is a forecast, and forecasts are what January inventory is made of.
Sourcing the reorder
A middle-of-the-run reorder only works if your supplier sells at size level rather than in fixed packs. Oversoles operates as a B2B wholesale platform with per-size availability, which is what makes this method practical rather than theoretical. Stock moves quickly during the autumn window, so work from the live collections rather than a fixed list.
