How to calculate sell-through rate

• ✓Sell-through % = net units sold ÷ (net units sold + ending on hand). With nothing but sales moving the stock, that is net units sold ÷ (opening stock + receipts). The overshirt: 540 ÷ 1,200 = 45.0%.
• ✓The base is the decision. Opening stock alone or receipts alone produces 125.0% or 62.5% on a style that really sold 41.7% to 44.2% of what it had.
• ✓Weekly sell-through says how fast stock is converting now; season-to-date says how much has gone. A receipt drops the cumulative figure without the style slowing.
• ✓Add units across sizes, doors and styles and divide once. Five doors averaging 62.0% sold 53.4% of their units.
• ✓Chase on full-price sell-through on core sizes with time to land a reorder; mark down when cover exceeds the weeks left; size the markdown to the surplus, by size.

One ratio, two ways to count it, two currencies

Sell-through compares what sold with what could have sold. Written out, the four forms a planner uses are:

• Sell-through % = net units sold ÷ (net units sold + ending on hand) × 100
• Sell-through % = net units sold ÷ (opening stock + receipts) × 100, when only sales moved the stock
• Value sell-through % = sales at original ticket ÷ (sales + ending stock, both at original ticket) × 100
• Units remaining = units available − net units sold

Take the overshirt that runs through this guide. It is a washed-twill style at a $64.00 ticket, received in two deliveries — 800 units in week 1 and 400 in week 5 — with no opening stock. By the end of week 8 it has sold 540 units and has 660 left. Sell-through is 540 ÷ (540 + 660) = 540 ÷ 1,200 = 45.0%. Because nothing but sales moved its stock, that is the same as 540 divided by the 1,200 units received, and the two forms agree. They stop agreeing the moment something other than a sale moves a unit — a transfer, a vendor return, shrink — and choosing between them is the subject of the next section.

Value sell-through answers the same question in money. For a single style at a single ticket it is identical to the unit figure: 540 units at $64.00 is $34,560, against $76,800 of stock received at $64.00, which is 45.0% again. It diverges as soon as more than one ticket is in the pool — several styles in a class, or one style whose remaining stock has been marked down. The rule that keeps it honest is that sales and remaining stock are valued at the same ticket, and the original ticket is the one that does not move. Value the sales at what they sold for and the stock at a reduced ticket, and the figure moves with every markdown whether or not a single extra unit sells.

The vocabulary drifts more than the arithmetic. “Sell-through”, “sell-thru” and “ST%” all mean this ratio. “Sell-in” is something else — the units a brand ships into a wholesale account, measured against what the account booked — and it is covered in the sell-in and bookings calculator. A sell-through figure on a report should carry three things beside it: the level it was taken at, the dates of the window, and whether it is units or value. Without them, two correct numbers can disagree by twenty points and nobody can say which is wrong.

Which base, and why it changes the answer

The numerator is rarely in dispute: units sold, net of the customer returns that went back into sellable stock. The denominator is where sell-through figures part company, and it is worth seeing how far apart they go on one ordinary period. Take a carryover chino in one group of doors. It opens the period with 400 units, receives 800, transfers 60 to another region, loses 10 to shrink, sells 540 and takes back 40 customer returns into stock. Net units sold are 540 − 40 = 500, and the period ends with 400 + 800 − 60 − 10 − 540 + 40 = 630 on hand. Five bases are in common use:

The first two are errors. Opening stock alone produces 125.0%, which cannot be a share of anything; receipts alone produces 62.5%, which credits this period with sales that came out of last period’s inventory. The third uses gross units sold, so the 40 units customers brought back still count as sold although they are back in stock and available to sell again. That leaves two defensible bases, 41.7% and 44.2%, and the gap between them is exactly the 70 units that left by transfer and shrink.

The working definition is net units sold ÷ (net units sold + ending on hand), because it measures the stock that was actually available to sell at that location for the whole window. Units transferred out were sold — or not — somewhere else, and they belong in the receiving location’s base. Units lost to shrink were never going to sell. Charging either to the sending location makes it look slower than its customers made it. Opening stock plus receipts is the right base for a different question: how much of what this location bought or was sent has turned into sales, including what it lost or gave away. That is a buying and loss question, and it is worth asking, but it is not the selling-rate question chase and markdown decisions run on.

At the level of a style across the whole business the two converge, because transfers between doors net out: a unit that leaves one door arrives at another. What remains between them is shrink and returns to vendor, which leave the business entirely. For a style-level total the difference is usually small; at door level, where transfers are the point, it can be large.

The same discipline applies to the numerator. Net units sold means gross units sold less the customer returns that went back into sellable stock. Returns that could not be resold — damaged, worn, out of season — are not netted back into stock and so do not come out of the units sold either; they are a loss line, handled in how to plan for returns. A channel with heavy returns, such as ecommerce, has a gross sell-through well above its net one, and the gap closes only as the return window passes, so its figure for the latest weeks is provisional for as long as the return window runs.

Seven steps, in order

Step four is the division. The other six decide whether the division describes what happened on the floor — the wrong window, a receipt dated by its purchase order or a door average gives a correct calculation of the wrong numbers.

1. 1

   Fix the unit and the window

   Decide what is being measured — style-color, size, door, channel or class — and over which dates. Write both down beside the figure. A sell-through without its level and its window cannot be compared with anything, including itself a week later.

2. 2

   Count what was available to sell

   Opening stock at the start of the window plus every receipt that became sellable inside it. Date receipts by when they reached the floor or the sellable pool, not by the purchase order or the ship date. Leave units that never belonged to this location or this window out of the base.

3. 3

   Count net units sold and ending on hand

   Net units sold are gross units sold less customer returns put back to stock. Ending on hand is the stock left at the close of the window. Note transfers out, vendor returns and shrink separately — they are the reason the two possible bases differ.

4. 4

   Divide on the right base

   Sell-through % = net units sold ÷ (net units sold + ending on hand) × 100. When nothing but sales moved the stock, that is the same as net units sold ÷ (opening stock + receipts). Never divide by opening stock alone or by receipts alone.

5. 5

   Compute period and cumulative together

   The week’s sell-through says how fast the stock on hand is converting now; the season-to-date figure says how much of everything received has gone. Read them side by side, because a receipt lowers the cumulative figure while the weekly figure barely moves.

6. 6

   Split it before you read it

   Separate full-price units from promotional and markdown units, then break the total out by size and by door or channel, summing units at every level rather than averaging percentages. Most decisions live in the split, not in the total.

7. 7

   Read it against the plan, then decide

   Compare actual sell-through with the planned curve for the same receipt dates, convert the gap to units, and check forward cover against the weeks left to sell. Ahead at full price on core sizes with time to land a reorder is a chase; behind with more cover than weeks is a markdown; everything else is a hold or a re-allocation.

Run the seven steps on the overshirt at the end of week 8. The unit is the style-color across all doors, and the window is weeks 1 to 8 (step one). It had no opening stock and received 1,200 units, the second delivery dated to the week it reached the floor (step two). It sold 540 net units and closed with 660 (step three). Sell-through is 540 ÷ 1,200 = 45.0% (step four), and the week-8 figure is 35 ÷ 695 = 5.0% (step five). Of the 540, 470 sold at full price, and the sizes S and M are running ahead of the rest (step six). Against a plan of 50.0% at week 8 it is 60 units behind, with 16.5 weeks of cover and 8 weeks left to sell, which points at a markdown sized to the surplus in the large sizes (step seven). Each step uses something the one before it established, which is the reason for the order.

Period against season-to-date sell-through

Sell-through can be taken over any window, and two windows do almost all of the work. Period sell-through — usually weekly — divides that week’s sales by the stock available in that week: opening stock plus any receipt that landed. Cumulative, or season-to-date, sell-through divides everything sold since the style launched by everything received since then. They answer different questions. The weekly figure says how fast the stock in hand is converting now; the cumulative figure says how much of the investment has come back. Here is the overshirt, week by week:

Three things are worth reading off it. First, the cumulative figure climbs steadily through week 4, to 40.6%, then drops to 33.3% in week 5. Nothing went wrong in week 5: the style sold 75 units against 70 the week before. The 400-unit second delivery entered the base, and the cumulative figure fell because its denominator rose by half. A report that shows only the cumulative line will make every second delivery look like a stall.

Second, the weekly figure tells a different and more worrying story. It runs between 11.3% and 12.8% through the first four weeks and then falls — 8.6%, 7.5%, 6.1%, 5.0% — as the share of stock converting each week declines and, from week 6, the units sold each week fall too. The cumulative figure is still rising through weeks 6 to 8, from 38.3% to 45.0%, which can read as momentum. It is not: it is a running total, and running totals rise as long as anything sells. The weekly rate is the one that shows the style slowing.

Third, the two figures are linked to cover. A weekly sell-through of s means the stock left after the week would last (1 − s) ÷ s more weeks at that week’s rate. At week 8, 5.0% converts to 660 ÷ 35 = 18.9 weeks of supply. That identity is why a falling weekly sell-through and a rising weeks-of-supply figure are the same news.

Longer windows sit between the two. A four-week or monthly period sell-through is the usual read in a merchandise plan, because it lines up with the planning calendar and smooths a single week’s noise. For the overshirt, weeks 5 to 8 opened with 475 units, received 400 and sold 215, a period sell-through of 215 ÷ 875 = 24.6%. Whatever the window, the rule is the same: opening stock at the start of it, plus receipts inside it, against net units sold inside it — and the window written beside the figure. A weekly 5.0%, a monthly 24.6% and a season-to-date 45.0% are all correct for the same style in the same week.

Units against value across a class

One style at one ticket has one sell-through. A class has at least three numbers that all claim to be its sell-through, and they can sit nine points apart. Take a three-style class at week 8: a core crew tee at $28.00, the overshirt at $64.00 and a quilted jacket at $148.00.

Unit sell-through for the class is 2,490 ÷ 4,800 = 51.9%. Value sell-through at original ticket is $107,160 ÷ $249,600 = 42.9%. The simple average of the three style percentages is 43.3%. The unit figure is high because the tee, which sells fastest, is also the deepest buy; the value figure is low because the jacket, which sells slowest, carries the highest ticket. The simple average is not a measure of anything, and lands near the value figure by coincidence.

Which figure to read depends on the decision, not on which one looks better. Unit sell-through is the operational read — it tells the warehouse, the allocators and the stores how much physical stock is converting, and it is what a size curve or a door allocation is checked against. Value sell-through is the financial read — it tells the planner how much of the money committed to the class has come back at ticket, and it is what the open-to-buy and the markdown budget care about. A class report should carry both, side by side, labelled, and should never switch between them from one week to the next.

The same arithmetic applies one level up. A department’s sell-through is its classes’ units, or values, added and divided once — never the average of the class percentages, and never a mixture of a unit figure for one class and a value figure for another. Where the classes inside a department sell at very different tickets, unit and value sell-through will diverge for the same reason they diverge here, and the gap between them is itself information: a value sell-through running well below the unit figure says the expensive end of the range is where the stock is sitting.

Ranking styles inside a class follows the same logic. A list sorted by sell-through percentage puts a small buy that sold 90 of 100 units above a deep one that sold 900 of 1,200, though the second sold ten times the units and may matter far more to the season. Sort by the gap to plan in units, or show the units beside the percentage, and the list reads in the order the decisions need to be taken.

Full-price against total sell-through

A unit sold on a 30%-off weekend is a unit sold, and it counts in total sell-through. It says much less about demand at the ticket. Of the overshirt’s 540 units, 470 sold at the $64.00 ticket and 70 sold on a 30%-off promotion at $44.80. Full-price sell-through is 470 ÷ 1,200 = 39.2%; total sell-through is 540 ÷ 1,200 = 45.0%. The promotional units are 70 ÷ 540 = 13.0% of the units sold and almost six points of the headline.

Chase and depth decisions read full-price sell-through; clearance decisions read the total. A reorder is a bet that the style will sell more units at the ticket, and the evidence for that bet is the units that sold at the ticket. A markdown decision is about whether the stock on hand will be gone by the exit date at whatever price it sells, and every unit that leaves counts toward that. A style that is strong only on promotion is telling the buyer its ticket is too high or its promotion is too generous, and buying more of it at the ticket is the wrong response to either.

The split needs a clean definition of full price. A permanent markdown resets the ticket, so units sold after it are not full-price units even if no promotion is running. A promotion is a temporary price, and units sold during it are promotional. Employee and loyalty discounts are usually small enough to be left in full-price units, provided the treatment is consistent. Where the price history is not kept by unit, the next-best split is by week: weeks with a promotion live against weeks without. In value terms the overshirt’s realised sales were 470 × $64.00 + 70 × $44.80 = $30,080 + $3,136 = $33,216, against $34,560 at original ticket for the same 540 units; the $1,344 difference is the promotional cost, and it belongs in the markdown and promotion budget rather than in the sell-through figure.

Size-level sell-through

A style-level sell-through averages its size run, and the decisions are mostly made inside the run. The overshirt was bought on a size curve of 8% XS, 22% S, 30% M, 25% L and 15% XL. Here it is at the end of week 8, with forward cover calculated on a forecast of 40 units a week split by each size’s share of units sold so far:

The style reads 45.0%. Its sizes read from 22.2% to 56.8%. S and M are selling well ahead of the style and ahead of the share of the buy they were given — S was 22.0% of the buy and is 27.8% of sales, M 30.0% of the buy and 37.0% of sales — while L and XL are behind on both. XS is slow, but it was a small buy and holds only 66 units. The style’s 45.0% hides a style that is two things at once: tight in the core sizes and long in the large ones.

Size-level sell-through is summed the same way as everything else: the style figure is the units sold across sizes over the units available across sizes, never the average of the five size percentages, which here would be 41.2%. And it has to be read with care about availability. A size that sells out stops selling, so its sell-through caps out while its demand does not. Shares of sales measured over weeks when a size was out of stock understate that size’s demand, and a forward split of the forecast based on them understates it again. Reading size sell-through over in-stock weeks, or checking whether any size has broken before trusting the shares, is what separates a curve problem from an availability one.

The weeks-of-cover column is where the decisions come from. At 40 units a week, S covers 10.3 weeks and M 10.8 weeks — more than the 8 weeks left, so neither needs chasing yet, though both will be the first to break. L covers 20.3 weeks and XL 47.3: on the current split, L and XL between them hold about 225 of the 340 units the style is projected to carry past its exit. A markdown on the whole style would spend margin on S and M units that are selling at the ticket; two-thirds of the surplus is in two sizes.

For next season, the gap between share of buy and share of sales is the input to the size curve, once it has been corrected for weeks out of stock. The mechanics of turning sales by size into a buy by size are in how to calculate size curves and the size curve calculator; the reason tail sizes are sometimes bought deeper than their sales share — so that every door can present the run — is worked through in minimum buy depth for a size run. A tail size planned to sell through slower for that reason should carry a lower planned sell-through from the start, so that it is not read as a failure.

Door-level and channel-level sell-through

Door-level sell-through is where the most common aggregation error in the subject lives. Take a knit polo allocated to five doors:

The five doors sold 534 of the 1,000 units they received: 53.4%. The average of their five percentages is 62.0%. The difference is 8.6 points, and it comes entirely from weighting: doors D and E, at 80.0% and 90.0%, hold 100 units between them, and in a simple average each counts as much as door A with its 400. Add the units sold across the doors, add the units available, and divide once. A store-percentage average is not a rougher version of the right answer; it is the answer to a different question — how the typical door is doing, regardless of size — and that question is rarely the one being asked.

Read as a set, the door figures carry their own decisions. D and E are nearly sold out on small allocations; they are likely broken in the core sizes and losing sales, and the question is whether they were under-allocated or simply small doors with a few strong days. C, at 30.0% on 200 units, is the door holding the surplus. Before any markdown is taken on the polo, a transfer of part of C’s stock to D and E may sell more of it at the ticket — provided the transfer cost and the transit time leave enough selling weeks at the other end. How doors are graded, how a reserve is held back for exactly this, and when a transfer is worth it are covered in how to allocate merchandise.

Small doors need patience. A door that received 40 units moves 2.5 points of sell-through with every unit sold, so its figure is noisy from week to week. Read small doors over longer windows, or in groups of similar doors, before acting on a single reading. The same applies to a new style in its first week, when a handful of units can produce a figure that means nothing.

Channels follow the same rules with one extra wrinkle. Ecommerce usually sells from a pooled stock rather than an allocation, so its base is the stock assigned to it — or, where stores fulfil online orders, the units it actually drew — and its numerator has to be net of returns that come back weeks later. A channel comparison that sets store sell-through on net units against ecommerce sell-through on gross units, or a store base against a pooled one, compares two different things. State the base for each channel, and add channels to a total by summing units, exactly as with doors.

When high sell-through is the problem

Sell-through has a ceiling of 100%, and demand does not. Once a style, a size or a door runs out, it stops selling, and its sell-through stops rising while its customers keep arriving. The figure records the units that sold and is blind to the ones that could have. A very high sell-through early in a season is as much a warning as a low one: it usually means the buy, the allocation or the size curve was too thin, and that sales were lost in the weeks the stock was not there.

Door E in the polo example sold 36 of its 40 units. If those 36 went in the first three weeks of an eight-week window, its rate was 36 ÷ 3 = 12 units a week, and for most of the remaining five weeks it had too little left to sell at that rate. Its 90.0% describes the stock it had; the demand it saw is a separate number that the sell-through figure cannot produce. The same is true of a size: the overshirt’s S and M hold barely more cover than the 8 weeks left, so they are the sizes that will break first if the forecast is light, and in any week either size is out, its share of sales understates its share of demand.

Two companion measures cover the blind spot. In-stock rate — the share of selling weeks, or door-weeks, in which the style or size was available — says how much of the window the sell-through was actually measured over. Rate of sale over in-stock weeks only — units sold divided by the weeks the item was in stock — estimates what it would have sold with stock behind it. A style with a high sell-through and a low in-stock rate was under-bought or under-allocated; a style with a high sell-through and a high in-stock rate sold well on a buy that was about right.

This is also why sell-through alone cannot set next season’s depth. Buying the depth that sold this season repeats this season’s stockouts; the depth has to be built from demand, which means correcting the units sold for the weeks without stock. Where the planning team has no way to see lost sales directly, the in-stock correction is the honest approximation, and it should be labelled as one.

Sell-through, weeks of supply and stock-to-sales

Sell-through is one of four stock-and-sales ratios that move together when the plan is right and diverge when it is not. Each one has a different base and a different time frame, and each one answers a different question. Sell-through looks backward at what has gone; weeks of supply and forward cover look forward at how long what is left will last; the stock-to-sales ratio sets the stock a period should open with; and inventory turn measures how hard a year’s inventory worked.

The links are arithmetic, not analogy. For one week, weeks of supply at that week’s rate is (1 − weekly sell-through) ÷ weekly sell-through: the overshirt’s 5.0% in week 8 is 18.9 weeks. Over a four-week average of sales — 75, 60, 45 and 35 units, or 53.75 a week — the same 660 units are 660 ÷ 53.75 = 12.3 weeks of supply. On the 40-unit weekly forecast the season ahead is expected to run at, they are 660 ÷ 40 = 16.5 weeks of forward cover. Three cover figures for one stock position, depending on which rate divides it. The weeks of supply calculator runs the trailing version; the markdown triggers in the markdown plan guide read the forward one, because on a declining style the trailing rate overstates demand and flatters cover.

The stock-to-sales ratio connects through the period. For weeks 5 to 8 the overshirt opened with 475 units and sold 215, a stock-to-sales ratio of 475 ÷ 215 = 2.21. Had nothing been received, period sell-through would simply be the reciprocal, 1 ÷ 2.21 = 45.3%. But 400 units landed, so period sell-through is 1 ÷ (stock-to-sales ratio + receipts ÷ sales) = 1 ÷ (2.21 + 1.86) = 24.6%, the same 215 ÷ 875 computed directly. The stock-to-sales ratio calculator and the stock-to-sales glossary entry cover how the ratio sets opening-stock targets; the point here is that a plan built on stock-to-sales ratios and a receipt plan implies a sell-through for every period, which is the plan curve the actual figure is read against.

Inventory turn is the annual cousin. A style that sells through quickly and is replenished turns its inventory many times; a seasonal style bought once and sold through to its exit turns roughly once over its life, whatever its sell-through rate. That is why a seasonal range is managed on sell-through to an exit date and a replenishment range on weeks of supply and turn — the measures fit how the inventory flows, and the inventory turnover calculator belongs to the second.

Reading it against receipt timing

Sell-through has receipts in its denominator, so anything that moves a receipt moves the figure. A delivery that lands early pulls sell-through down before the units have had a chance to sell. A delivery that lands late flatters it — the style looks as if it is selling through its stock, when it is selling through the part of its stock that arrived. Here is the overshirt with its 400-unit second delivery landing on time in week 5, and three weeks late in week 8, with sales held identical in both:

On time, cumulative sell-through runs 33.3%, 38.3% and 42.1% through weeks 5 to 7. Late, it runs 50.0%, 57.5% and 63.1%. By week 7 the late version reads 21 points ahead on the same 505 units sold, and a buyer looking only at the figure would see a style that is outselling its plan. A late receipt flatters sell-through until it lands and then punishes it: in week 8 both versions read 45.0%, and the late version now has 400 fresh units and three fewer weeks to sell them in.

Holding sales identical is the generous assumption. In practice a late delivery usually means the core sizes broke before it arrived, so some of those weeks’ demand was lost rather than delayed, and the late version would have sold fewer units, not the same. The sell-through figure cannot see lost sales at all — it only sees units that sold — which is why a late style can read well and still have missed its plan in dollars.

Three rules follow. Date every receipt by when it became sellable: on the floor in a store, in the sellable pool for ecommerce. A receipt booked on its purchase-order date or its ship date sits in the base for weeks before any customer can buy it. Second, when a delivery moves, rebuild the plan curve on the actual date before comparing against it; otherwise the variance is mostly the delivery, not the demand. Third, read late styles on weeks on floor rather than calendar weeks — what the style sold in its first four weeks of availability against what the plan expected in its first four weeks — which strips the timing out.

Receipt timing is a plan of its own, with its own failure modes, and the phasing of deliveries across a season is covered in how to plan receipt flow. Where styles land in drops, each drop has its own launch week, and a class sell-through that mixes a drop in its first week with one in its eighth averages two different stages of life. Reading by drop, or by weeks on floor, keeps like with like.

Reading it against a plan curve

A sell-through figure on its own is a fact without a judgement. 45.0% at week 8 is ahead for a heavy-outerwear style with twenty weeks to sell and far behind for a fashion drop meant to clear in six. The judgement comes from comparing it with what the plan expected for that style, at that point, given its receipts. The planned sell-through curve is built from the style’s own sales plan and receipt plan, week by week: planned cumulative units sold ÷ planned cumulative units received. It is not imported from another business, another category or a published table, because none of those had this style’s receipts or this season’s length.

At week 4 the overshirt is 15 units behind a plan of 340 — 40.6% against 42.5%, close enough to watch rather than act. By week 8 the gap is 60 units, 45.0% against 50.0%, and the weekly rate is still falling. Converting the gap to units is the important habit. A five-point gap on a 1,200-unit buy is 60 units; on a 12,000-unit buy it is 600. The percentage is the same, and the size of the problem is not.

A flat target — “sell 70% by week 12” across a whole range — fails for the reasons the earlier sections set out. Styles land at different times, so a flat target compares a style in its third week with one in its tenth. Styles are bought at different depths relative to their demand, deliberately, so the same week can be early in one style’s curve and late in another’s. And categories have different shapes: a basic sells at a steady rate, a fashion style front-loads, a gifting style back-loads. The plan curve absorbs all three because it is built from the same plan that set the buy.

Where no style-level plan exists — a new style with no like-for-like history, or a range planned only at class level — the next-best reference is the curve of a comparable style from a previous season, restated onto this season’s receipt dates. That is a proxy and should be labelled as one. What should not be used as a reference is last year’s sell-through on the same week number with no adjustment for receipts, which compares this year’s style with last year’s delivery schedule.

Building the planned curve, and planning the exit

The plan curve is not a separate forecast. It falls out of two things the merchandise plan already holds for the style: its phased sales plan, week by week, and its receipt plan, dated to when each delivery becomes sellable. Divide planned cumulative sales by planned cumulative receipts for each week and the curve is done. For the overshirt, the plan put 340 units of sales in weeks 1 to 4 and 260 in weeks 5 to 8, against 800 units received in week 1 and 400 in week 5. That gives 340 ÷ 800 = 42.5% at week 4 and 600 ÷ 1,200 = 50.0% at week 8.

Because the curve is built on the receipt plan, it dips when a planned delivery lands, exactly as the actual figure does. With 70 units planned for week 5, the overshirt’s planned cumulative sell-through falls from 42.5% at week 4 to 410 ÷ 1,200 = 34.2% at week 5. Read against that curve, the actual drop to 33.3% in week 5 is a variance of less than a point rather than a seven-point collapse. This is the practical case for building the curve from the plan instead of drawing a smooth line: a smooth line puts every delivery’s dip into the variance.

The far end of the curve is the planned exit sell-through: the share of the units the style was bought at that is planned to have sold by its exit date, with the rest going to an outlet, a clearance event or a carryover decision. As an illustration only, a 1,200-unit buy planned to leave 120 units for the outlet has a planned exit sell-through of 1,080 ÷ 1,200 = 90.0%. That figure is a planning decision, taken when depth is set, and it ties the buy, the markdown budget and the exit together: the deeper the buy relative to expected demand, the lower the exit sell-through at full price, and the more markdown the plan has to budget to clear the difference.

The projection at week 8 is then a comparison of two exit figures. On its current forecast the overshirt closes the season at 71.7%; against an illustrative 90.0% exit plan, that is 1,080 − 860 = 220 units the plan expects to sell and the forecast does not, or 220 × $64.00 = $14,080 at original retail. That is the size of the decision the markdown plan has to make, expressed in the same units the buy was made in. A planned exit that was never written down leaves this comparison with nothing on one side, and the in-season markdown conversation turns into an argument about what the style was supposed to do.

The launch read: what two weeks of sell-through can and cannot say

The most valuable sell-through reading is also the least reliable one. In the first week or two after a style lands, almost every lever is still open — a reorder can still arrive in time, an allocation can still be rebalanced, a markdown is still a long way off — and the evidence is a few hundred units at most. The overshirt sold 175 of its first 800 units in two weeks, a cumulative sell-through of 21.9%, against a plan of 180, or 22.5%. Five units behind on 800 is not a signal; it is noise.

An early read is a question about rate, not about level. Two weeks of sell-through cannot say where a style will finish, but they can say whether its weekly rate is in the range the plan assumed, and in which sizes and doors it is selling. That is enough to make the cheap decisions early — rebalance a reserve, hold back a transfer, flag a style for a closer look next week — and to defer the expensive ones until the evidence is better. A chase placed on two weeks of sell-through is a bet on a small sample, and the size of the bet should reflect that.

Three checks make a launch read more trustworthy. First, confirm the style was actually available everywhere it was planned to be — a style that reached half its doors a week late has a launch sell-through that measures the delivery, not the demand. Second, read full-price units only, because launch weeks often coincide with marketing or promotional events that pull sales forward. Third, compare like with like: the style’s first two weeks against the first two weeks of comparable styles from previous seasons, on their own receipt dates, rather than against calendar weeks.

The reason to take the early read at all, despite its noise, is that the cost of waiting rises every week. A reorder that could have landed with eight weeks to sell lands with four; a transfer that could have caught the peak misses it; a style that needed a small markdown early needs a deep one late. A planning team that waits for sell-through to be certain will usually find it certain about a decision that has already expired.

What sell-through should trigger: chase, hold, markdown

Sell-through is read so that something can be decided. The decision rules work best when they are agreed before the season, so that the weekly meeting reads them off rather than argues them. Four readings cover most cases:

Work the overshirt at week 8. It is behind plan, so a chase is off the table. Forward cover is the test for a markdown. On the 40-unit weekly forecast for the eight weeks it has left, it is projected to sell 8 × 40 = 320 more units, closing the season at (540 + 320) ÷ 1,200 = 71.7%, with 660 − 320 = 340 units left over. Its 660 units on hand are 16.5 weeks of cover against 8 weeks to sell. The markdown is sized to the 340 projected to remain, and placed where they sit: the size table puts about 225 of them in L and XL, while S and M, at 10.3 and 10.8 weeks of cover, are close to selling through at the ticket. A whole-style markdown would spend margin on core-size units that were going to sell anyway.

How deep the markdown goes, when it is taken and what it costs are the markdown plan’s questions rather than sell-through’s — the depth ladder, the exit date and the budget are worked through in how to build a markdown plan, the price and margin of one cut in the markdown calculator, and the extra units a cut has to sell to hold margin dollars flat in the break-even markdown calculator. What sell-through contributes is the size of the surplus and where it sits.

A chase is the mirror image, with one extra condition: time. A style running ahead of plan at full price on its core sizes justifies more stock only if a reorder can land while there are still weeks to sell it, so the chase decision is a sell-through reading checked against lead time. A strong style with a twelve-week reorder lead time and eight weeks left in its season is not a chase; it is a lesson for next season’s depth. The reading on why the decision has to come early, while the evidence is still thin, is in in-season chase: reading sell-through before it is too late.

Every one of these decisions changes the rest of the season’s numbers — a markdown moves the sales and markdown lines, a chase adds receipts, a transfer moves stock between locations — so a sell-through decision is followed by a reforecast of the periods it touches, and the open-to-buy those periods still have. That sequence is covered in how to reforecast in-season.

How sell-through flows into open-to-buy

Sell-through is usually read style by style, but its consequence lands in the merchandise plan as stock. A unit that was planned to sell and did not is a unit the next period opens with, and every unit of opening stock above plan is a unit the open-to-buy for that period no longer has room for. A sell-through gap is an open-to-buy problem in disguise, and the conversion runs in units first and retail dollars second.

The overshirt’s 60-unit gap at week 8 is 60 × $64.00 = $3,840 of retail stock the plan expected to have sold by now. If nothing changes, that stock rolls into the following weeks’ opening position, and the receipts planned for those weeks — for this style or for whatever was meant to take its place on the floor — have $3,840 less room in them. The open-to-buy identity makes the link exact: planned closing stock plus planned sales plus planned markdowns, less opening stock, less stock already on order. Raise the opening stock and, with everything else held, the open-to-buy falls by the same amount.

The three responses to a gap each move a different line of that identity. A markdown moves stock out by moving the markdown line up, and spends margin to do it. A transfer moves stock between locations without changing the total, which helps only if the receiving location has the demand. A cut or a push of a later receipt moves the on-order line down or later, which is often the cheapest response and the one that expires first, because a receipt can only be cancelled before it ships. Reading the gap early is what keeps the third option open.

The arithmetic of the identity is in how to set open-to-buy and the open-to-buy calculator; the weekly view of the same stock — opening, intake, sales and closing — is the WSSI, read in how to read a WSSI. Sell-through tells the planner which styles created the gap; the WSSI and the open-to-buy tell them which weeks it lands in and how much buying room it costs.

Returns, transfers and shrink

Three stock movements that are not sales or receipts account for most of the disagreements between two people calculating the same sell-through. Each has a defensible treatment, and the treatment matters less than holding it constant.

Customer returns come out of units sold when they go back into sellable stock, so the numerator is net units sold. In a channel with a long return window, the latest weeks’ net figure is provisional until the window closes, and a style’s sell-through can drift down after the fact as returns arrive. Report recent ecommerce sell-through as provisional, or on a lag, rather than letting it be revised silently. Returns that cannot be resold stay counted as sold and are written off as a loss, outside the sell-through figure.

Transfers move units between locations without leaving the business. On the net-units-sold-plus-ending-on-hand base, a unit transferred out simply leaves the sending location’s base and joins the receiving location’s, which is the treatment that measures each location on what it actually had to sell. Transfers in arrive like a receipt, dated to when they became sellable. At style level across the business, transfers cancel.

Shrink and returns to vendor leave the business. On the recommended base they leave the denominator, because units that are gone cannot be sold; on the opening-plus-receipts base they stay in it as if unsold. Where shrink is booked from a physical count rather than continuously, a count can change a location’s ending on hand overnight and move its sell-through with it, so a sudden jump in a door’s figure in a count week is checked against the count before it is read as demand. The chino earlier in this guide shows the size of the effect: 41.7% on opening plus receipts, 44.2% on net units sold plus ending on hand, with the 70 units of transfer and shrink the only difference.

Carryover styles and continuity items

A style that sells across more than one season has more than one sell-through, and the one being quoted has to be named. The carryover chino from earlier in this guide was launched last season. Last season it received 1,600 units, sold 1,190 net and lost 10 to transfers and shrink, closing with 400. This period it opened with those 400, received 800 more and sold 500, closing with 630 after 70 units of transfers and shrink. Three figures describe it:

• Last season: 1,190 ÷ (1,190 + 400) = 74.8%
• This period: 500 ÷ (500 + 630) = 44.2%
• Life to date: 1,690 ÷ (1,690 + 630) = 72.8%

Life-to-date sell-through on a carryover style is dominated by its history: the chino’s 72.8% is mostly last season, and it would take a very bad period to move it far. The period figure is the one that says how the style is selling now, and it is the one a decision on this season’s depth should read. The life-to-date figure is the one that says whether the style, over its whole life, has earned its place in the range.

Continuity items take this to its limit. A replenished style never has a season end, so a cumulative sell-through on it rises toward a figure that reflects only how long it has been in the range. What governs a continuity item is its weekly rate against its replenishment plan: weekly sell-through, weeks of supply against the reorder point, and in-stock rate. The reorder point, minimum and maximum that turn that rate into receipts are worked through in how to plan a replenishment program.

The trap is mixing the two in one report. A class holding seasonal styles read cumulatively and continuity styles read the same way will show the continuity styles as either very strong or very weak depending on how long they have been carried, and neither reading means anything. Read seasonal styles cumulative to an exit date and continuity styles weekly against their replenishment plan, and report the two groups separately.

Sell-through at a wholesale account

For a brand that sells wholesale, two different numbers get called sell-through, and keeping them apart avoids a recurring argument. Sell-in is what the brand shipped to the account against what the account booked. Sell-through at the account is what the account’s customers bought, divided by what the account had to sell — its opening stock plus what it received —the same formula as everywhere else in this guide, calculated on the account’s stock rather than the brand’s.

The brand sees account sell-through only when the account shares its point-of-sale and stock data, and the data arrives on the account’s calendar, at the account’s level of detail and with the account’s definitions. Before comparing two accounts, check that they report on the same base — units received or opening plus receipts — and the same treatment of returns and transfers between their own doors. An account that reports sell-through on receipts alone will look stronger than one that includes its carryover stock, for the reason the chino showed.

What a brand does with account sell-through is usually a conversation rather than a markdown: a reorder from the account if the style is running ahead, a request for markdown support or a return if it is running behind, and a depth decision for next season’s line. The arithmetic of booking attainment and fill rate on the sell-in side is in the sell-in and bookings calculator, and planning one buy against booked wholesale and forecast direct demand is in how to plan wholesale and DTC together.

Reviewing a sell-through report: six questions

A sell-through report arrives as finished percentages. Six questions, asked in order, establish whether those percentages mean what they appear to. Each one maps to a specific way the figure can be right as arithmetic and wrong as a reading.

1. What is the base? Opening stock plus receipts, or net units sold plus ending on hand — and is it the same on every row? Any figure above 100% is a base error.
2. What is the window? Week, period, season to date or life to date, stated on the report, with the dates. A weekly and a cumulative figure on the same row are both useful; an unlabelled one is not.
3. How were the totals built? Recompute one class or door-group total as summed units over summed units. If it differs from the figure on the report, the report averaged percentages.
4. Units or value, and at which ticket? Value figures should carry sales and remaining stock at the same ticket, normally the original one, so that a markdown does not lift the figure on its own.
5. When did the receipts land? Check that receipts are dated to when they became sellable, and flag any style whose delivery moved, so its figure is read against a curve rebuilt on the actual date.
6. What was the plan, in units? Each style’s figure needs its planned sell-through beside it and the gap converted to units, so that a five-point miss on a small buy and on a large one are not treated as the same problem.

None of the six needs a model. Each needs the report to carry its own definitions — base, window, aggregation, currency, receipt dating, plan — so that the reader can check the figure instead of trusting it. A report that cannot answer them is a list of percentages, and the decisions taken from it inherit whatever definitions happened to be used in the export.

Nine ways a sell-through figure misleads

Each produces a number that divides correctly. None shows up as a spreadsheet error; each shows up as a chase that should not have been placed, or a markdown taken in the wrong sizes or the wrong week.

What changes about sell-through in each vertical

The formula does not change. What changes is the unit it is read at, the window it is read over, and which units belong in the base — so a sell-through figure is compared within a vertical, on one definition. Apparel is the reference case.

• Sell-through calculator — the percentage and the units remaining →
• Weeks of supply calculator — how long the stock left will last →
• Stock-to-sales ratio calculator — opening stock against period sales →
• Markdown calculator — the price and margin of a single cut →
• Size curve calculator — sales by size into a buy by size →
• How to build a markdown plan — the triggers sell-through feeds →
• How to reforecast in-season — re-projecting the season after a decision →
• How to allocate merchandise — doors, reserves and transfers →
• How to plan receipt flow — the receipt dates in the denominator →
• How to calculate size curves — reading sales by size →
• Retail math formulas — the wider formula reference →
• Sell-through rate — glossary definition →
• Forward cover — glossary definition →
• Stock-to-sales ratio — glossary definition →
• Markdown — glossary definition →
• Size curve — glossary definition →

Frequently asked questions

How do you calculate sell-through rate?

Sell-through rate = net units sold ÷ (net units sold + units on hand at the end of the period) × 100. When nothing but sales moved the stock, that equals net units sold ÷ (opening stock + receipts). In this guide’s example, an overshirt that has sold 540 units and has 660 left has a sell-through of 540 ÷ (540 + 660) = 540 ÷ 1,200 = 45.0%.

Should sell-through be calculated on beginning inventory or on units received?

On both together: opening stock plus every receipt that became sellable in the period, or equivalently net units sold plus ending on hand. In this guide’s carryover chino, 500 units sold reads 125.0% on the 400 units of opening stock alone, 62.5% on the 800 units received alone and 41.7% on the 1,200 units it actually had to sell. Measured as net units sold plus ending on hand, which leaves out 60 units transferred out and 10 lost to shrink, it is 500 ÷ 1,130 = 44.2%.

What is the difference between unit sell-through and dollar sell-through?

Unit sell-through counts units; dollar (value) sell-through weights each unit by its original ticket, with sales and remaining stock valued at the same ticket. They are equal for one style at one ticket and diverge across a class. In this guide’s three-style class, 2,490 of 4,800 units sold is 51.9%, while $107,160 of $249,600 at original ticket is 42.9%, because the slowest style carries the highest ticket.

What is the difference between weekly and season-to-date sell-through?

Weekly (period) sell-through divides one week’s sales by the stock available that week; season-to-date (cumulative) sell-through divides all units sold so far by all units received so far. In this guide’s example the overshirt’s week-8 sell-through is 35 ÷ 695 = 5.0%, and its season-to-date sell-through is 540 ÷ 1,200 = 45.0%. A receipt lowers the cumulative figure immediately, which is why the two are read together.

How do you calculate sell-through across several stores?

Add the units sold across the stores, add the units each store had available, and divide once. Never average the store percentages. In this guide’s example, five doors sold 534 of 1,000 units, a sell-through of 53.4%, while the simple average of their five percentages is 62.0%, because two small doors at 80.0% and 90.0% count as much as a door holding 400 units.

How is sell-through related to weeks of supply?

For a single week, weeks of supply at that week’s sales rate = (1 − weekly sell-through) ÷ weekly sell-through. In this guide’s example the overshirt sold 35 units in week 8 from 695 available, a weekly sell-through of 5.0%, and closed with 660 units: 660 ÷ 35 = 18.9 weeks of supply at that rate. Sell-through says how much has gone; weeks of supply says how long what is left will last.

What is a good sell-through rate?

There is no universal figure, and this guide offers none. A good sell-through is one at or ahead of the plan curve built from the style’s own sales plan and receipt dates, at full price. In this guide’s illustrative plan, the overshirt was planned to sell 600 of 1,200 units by week 8, a planned sell-through of 50.0%; it sold 540, or 45.0%, which is 60 units behind.

When should sell-through trigger a markdown?

When sell-through is behind plan and the stock on hand covers more weeks of forecast sales than the season has left. In this guide’s example the overshirt has 660 units on hand, a forecast of 40 units a week and 8 weeks left to sell: 16.5 weeks of cover against 8, with 320 units projected to sell and 340 projected to remain. The markdown is sized to the 340, by size, not to the whole style.