Syntha Lab RU

Size runs, stores and range in season

Whether the size run is intact, whether moving stock between stores pays and whether a new item takes sales from its neighbors.

For retail Planning and buying

Open interactively →

About the topic

By the start of the season the buy is done and the question changes: not “what to buy” but “what to do with what is on the shelves”. Which stores received the right sizes, where are the gaps, what is worth moving and what should stay put. The topic answers three questions: is the size run of a model intact, does moving stock between stores pay off, and does a new item add sales or take them from its neighbors.

Sizes are a topic of their own because demand by size is uneven and the shopper is not flexible. A size run is a “run” for a reason: people take their size or nothing. If a sneaker run has no 42, the shopper will not take a 41 or a 43. They leave, and the neighboring sizes sit until the end of the season and go out on markdown. That is why a hole in the core sizes costs more than their share of the buy: along with the lost pair go the add-on items and the visit itself.

The reasoning goes like this. First measure instead of guessing: take the share of each size in the sales of a model over the weeks when the size was in stock, and find the core, the sizes that together give about two thirds of sales. Then pick an action from cheapest to dearest: replenishment from the warehouse, transfer between stores, markdown. A transfer pays if the markdown avoided and the margin recovered exceed the cost of moving, adjusted for the chance of selling. Example: a pair costs RUB 7,000, a 30% markdown eats RUB 2,100, and moving costs RUB 150 per pair plus RUB 400 per shipment: six pairs move for RUB 1,300 against up to RUB 12,600 lost on markdown. Cannibalization asks the same question at range level: what counts is not what the new item sold but what it added beyond what it took from its neighbors.

Start with one fast-selling model. Pull sales and stock by size and store for four to eight weeks, remove the weeks when a size was out of stock and the weeks of markdown, and see which sizes form the core and where they are empty. The main traps: reading zero sales at zero stock as zero demand, taking a store’s last set, moving goods without counting the cost of the move, and measuring cannibalization without control stores.

Key ideas

Core sizes

The core is a few neighboring sizes that give most of a model’s sales. If sizes 42, 44 and 46 in a women’s run give 62% of sales together, their stock matters more than the other five. That is why run completeness is counted not by the number of sizes in stock but by the share of sales that the stock covers. In the “Size run” calculation the core threshold is a field (75% of sales by default): set it for your own model.

A hole costs more than its share

A size that makes 18% of sales takes no more than those 18% of the size’s own sales from the model when it is missing, but costs more than that: shoppers do not take a neighboring size and leave, and the add-on purchase in the same basket does not happen either. That is why the time to reorder or move stock is shorter for the core than for the edges.

Moving versus markdown

Moving is a way to “buy” a sale at the price of logistics instead of the price of markdown. If moving a pair costs RUB 150 and markdown takes away RUB 2,100, the gain per pair = (chance to sell at the receiver - chance to sell at the donor) x margin - moving cost. With a margin of RUB 2,100 and chances of 65% at the receiver and 25% at the donor that is 0.4 x 2,100 - 150 = RUB 690; if the receiver’s chance is only 30% against 25%, the gain of 0.05 x 2,100 = RUB 105 is below RUB 150 and moving loses money. A fixed fee per shipment makes small batches lose money too.

Minimum display set

A store needs a minimum for the item to look good and sell: for example, one pair of each core size. Only the stock above this minimum can be taken: otherwise the store becomes a warehouse of leftovers and the model stops selling on the spot.

Cannibalization

A new item seldom adds all of its own revenue. If the novelty sold 100 units and the category in test stores grew by 60 more units than in control stores, then 40 units were taken from neighbors. With a margin of RUB 3,000 on the novelty and RUB 2,500 on the neighbors the net effect is 100 × 3,000 − 40 × 2,500 = RUB 200,000, not RUB 300,000.

The edge of the size run is returned more often: one point costs money

The sizes XS and XXL are returned more often than the middle (34% and 38% against 24% for M). The margin per ordered unit in the best group is RUB 878, in the worst RUB 572: a gap of 35%. One point of returns costs RUB 21.9 per unit, so cutting the XXL returns from 38% to 30% (an exact measurement table, a photo on a model of the same size) gives RUB 14,003 on 80 units. It is cheaper than under-ordering the edge: a hole in the run costs sales, while a return costs only part of the margin.

Formula: The price of a return point = (m + return + handling + scrap share × cost) ÷ 100

What to do: First cut the returns of the edge sizes with description and measurements, then decide whether to shorten the edge of the run.

Think first: XXL returns fell by 8 points. Is the gain counted on the whole batch or only on the XXL size?

Answer: Only on XXL: 80 units × 8 points × 21.9 ₽ = 14,003 ₽.

A curve by sales understates sizes without stock

The rate of a size over its in-stock days shows demand: on illustrative data the true share of L is 31.1%, 65 units were lost.

Formula: Share = (sales ÷ in-stock days) ÷ sum over sizes

What to do: Take in-stock days by the store, not by the warehouse, and buy by the true curve.

How to work through it

  1. Pick one fast-selling model and a period: four to eight weeks without markdowns and without weeks when key sizes were out of stock.
  2. Work out each size’s share of sales and find the core: the sizes giving about two thirds of sales.
  3. Compare with stock: which core sizes are empty in which stores and how many sales that costs.
  4. Check replenishment from the warehouse first: it is cheaper than moving between stores.
  5. For surpluses in some stores and shortages in others, calculate the move: shipping cost, chance of selling, markdown avoided.
  6. Leave every store a minimum set of core sizes and do not take the last of it.
  7. Before launching a new item take test and control stores and measure how much it added and how much it took from neighbors.
  8. Two or three weeks later check the result: has the run recovered and has the markdown been avoided.

Where to start

Start with “Size-run completeness” on one fast-moving style, then “Does moving stock pay back” on its store stocks. Calculations of the topic in order: “Size-run completeness and core sizes”, “Does moving stock between stores pay back”, “Net effect of a new item: cannibalization”, “Size demand corrected for stock-outs”. See also (their primary topic is another one): “Size recommendation: payback of the tool”, “Size curve by store cluster and pack composition”, “Replenishment by size: what and how much to send to stores by size”, “Transfers between points: surplus, need and the gain against a discount”, “Swimwear and lingerie: returns by size run, hygiene write-offs and margin per ordered unit”.

Calculations in this topic

  • Size-run completeness and core sizes

    Finds the core sizes of a style, shows whether the run is intact, how many weeks of sales each size covers and which sizes to top up or redistribute.

    Whether a style’s size run is intact, which sizes to top up and which to redistribute.

  • Does moving stock between stores pay back

    Identifies donors and recipients by cover to the season end, the pieces to move, the gain per piece (with sale probabilities), the total result including fixed costs and the break-even transport cost.

    Whether moving surplus from one store to another pays after the season starts.

  • Net effect of a new item: cannibalization

    Compares the margin gained from a new item with the margin lost on neighboring items after launch (adjusted for the overall trend in control stores). Shows the net effect on the category and the cannibalization share.

    Whether a new item widens the category’s sales or takes them from neighbors.

  • Size demand corrected for stock-outs

    Restores the true size curve: divides the sales of each size by the days it was in stock. Shows which sizes are understated in the curve by sales, how many units were lost and how to split the purchase.

    What the true size curve is if some sizes were not in stock for the whole period, and how many sales were lost.

  • Size recommendation: payback of the tool

    Calculates whether an online size-recommendation tool (a widget, a measurement-based hint, a size chart) pays back: how many size returns it removes, what one return costs, the net effect per month and the payback of the launch cost. Shows the reduction in returns at which the tool breaks even.

    Whether a size-recommendation tool pays back and at which reduction in returns it breaks even.

  • Size curve by store cluster and pack composition

    Splits the buy across store clusters and sizes using each cluster’s past sales, picks the composition of a ready pack (prepack) for the cluster’s curve, rounds the buy to whole packs and shows how many pieces end up in the wrong sizes. Compares a pack per cluster with one pack for the whole chain, and picking to order with a ready pack.

    How many pieces of which sizes to send to each store cluster and which pack to choose.

  • Replenishment by size: what and how much to send to stores by size

    From the stock, the goods in transit and the sales of each “store × size” cell calculates the target stock in weeks of cover with a minimum display, the need rounded up to the pack multiple and distributes the limited warehouse across sizes: cells with the lowest cover first. Shows how much to send and what the warehouse lacks.

    What and how much to send to stores by size from a limited warehouse.

  • Transfers between points: surplus, need and the gain against a discount

    From the stock and selling speed of each point (a store or a warehouse) finds where there is a surplus above the required cover and where a shortage, picks transfers with lot multiples and compares the cost of moving with the discount that would have to be given to clear the surplus on the spot.

    Where to move surplus and whether moving pays better than a discount on the spot.

  • Swimwear and lingerie: returns by size run, hygiene write-offs and margin per ordered unit

    From the price, the cost, the channel commission, the logistics and a table of size groups (share of orders and share of returns) it counts the weighted return rate and the margin per ordered unit including returned goods that cannot be resold (the hygiene liner is removed, the item has been worn). It shows the loss caused by returns, returns in units, the return rate at which the margin falls to zero and the size groups that pull the margin down. Check the requirements for returning lingerie and swimwear and the list of items that cannot be resold after a return against the current sales rules; the share that cannot be resold and all default values are illustrative.

    How many returns the margin of a lingerie or swimwear item can bear, which size groups pull it down and how much goods is written off after a return.

Self-check questions

  • Which sizes of the model form the core and what share of sales do they give?
  • In how many stores is the core incomplete and how many sales does that cost per week?
  • Are weeks of markdown and weeks when a size was out of stock excluded from the calculation?
  • Can the hole be closed by replenishment from the warehouse instead of a move?
  • What does moving a pair cost once the fixed shipping fee is included?
  • What is the chance that the moved size sells in the receiving store?
  • What minimum set stays in the donor store?
  • Does the new item have control stores where it is absent?
  • Does the region need its own size curve or is the common one enough?

Knowledge check

Multiple-choice questions. The correct answer and the explanation follow each question.

  1. Sales of a model across six sizes: 50, 40, 28, 22, 12 and 8 pairs. The core is the sizes in descending order of sales until the cumulative share reaches 75%. How many sizes are in the core?

    • A. 3 sizes
    • B. 4 sizes
    • C. 2 sizes
    • D. 5 sizes

    Correct answer: B. 4 sizes. The total is 160 pairs. Cumulative share: 50 (31%), 90 (56%), 118 (73.75%), 140 (87.5%). Three sizes give 73.75%, which falls short of 75%, so a fourth is taken: the core is 4 sizes. Someone who rounds 73.75 up to “about 75” will pick three. Two sizes give only 56%. Five are needed if the threshold is 90% or higher, and ours is 75%.

  2. Moving pairs between stores: the margin of a pair is RUB 1,800, the chance to sell at the recipient is 70%, at the donor 30%, shipping is RUB 300 a pair plus a fixed fee of RUB 1,500 per shipment. What is the smallest batch that pays for the shipment?

    • A. 1 pair
    • B. 3 pairs
    • C. 4 pairs
    • D. 5 pairs

    Correct answer: C. 4 pairs. The benefit per pair = (0.7 − 0.3) × 1,800 − 300 = RUB 420. A batch of n pays off if 420 × n > 1,500, that is n > 3.57: the smallest batch is 4 pairs (3 pairs give RUB 1,260, below the fee). Three pairs appear if the RUB 300 per pair is forgotten (720 × 3 = 2,160 > 1,500). One pair appears if only the margin of 1,800 is compared with the fee. Five pairs is 1,500 / 300: the fee was compared with shipping alone.

  3. Which size group gives the lowest margin per ordered unit at returns of 34, 28, 24, 27, 33 and 38% for XS–XXL?

    • A. XS
    • B. M
    • C. XL
    • D. XXL

    Correct answer: D. XXL. The margin falls with returns, so the worst group is the one with the largest return rate: XXL, RUB 572. The best is M, RUB 878. Covered in: The edge of the size run is returned more often: one point costs money.

  4. Shelf price RUB 4,200, commission 18%, cost RUB 1,300, return RUB 150 and handling RUB 60, scrap 35%. How much does one point of returns cost per ordered unit?

    • A. 15.2 ₽
    • B. 21.9 ₽
    • C. 60 ₽
    • D. 13 ₽

    Correct answer: B. 21.9 ₽. What is lost is the margin on the kept unit, the return, the handling and the scrapped part of the cost: RUB 2,188 per unit when ρ grows by 100%, that is RUB 21.9 per point. The margin alone, RUB 15.2, is too little. Covered in: The edge of the size run is returned more often: one point costs money.

  5. XXL returns (80 units in the order) were cut by 8 points. What is the gain in margin?

    • A. 175,036 ₽
    • B. 175 ₽
    • C. 14,003 ₽
    • D. 640 ₽

    Correct answer: C. 14,003 ₽. 80 units × 8 × RUB 21.9 = RUB 14,003. It cannot be counted on the whole order (RUB 175,036): returns changed only for XXL. Covered in: The edge of the size run is returned more often: one point costs money.

  6. How to restore the real share of a size?

    • A. Take sales as they are
    • B. Take the stock in the warehouse
    • C. Take last year share
    • D. Divide sales by in-stock days and normalize

    Correct answer: D. Divide sales by in-stock days and normalize. The rate over in-stock days is not distorted by stock-outs. Covered in: A curve by sales understates sizes without stock.

  7. A size was in stock three days of 60. How to read its rate?

    • A. As reliable
    • B. Unreliable: take a neighbor size or an analog
    • C. Ignore the size
    • D. Double the purchase

    Correct answer: B. Unreliable: take a neighbor size or an analog. A rate over a few days is random. Covered in: A curve by sales understates sizes without stock.

Common mistakes

  • Mistake: Building the curve from weeks when a size was out of stock

    What it costs: Zero sales are read as zero demand and the core is understated

    How to avoid: Count only weeks with stock or use the share in sales at a full run.

  • Mistake: Taking a store’s last set

    What it costs: The store loses sales on the spot and the move does not pay

    How to avoid: Set a minimum core set and move only what is above it.

  • Mistake: Moving without counting the shipping fee

    What it costs: Small batches lose money because of the fixed fee

    How to avoid: Compare the cost of moving a batch with the expected markdown avoided.

  • Mistake: Measuring cannibalization on revenue without control stores

    What it costs: Seasonal growth is credited to the novelty

    How to avoid: Compare test stores with control stores in the same weeks.

  • Mistake: Carrying one region’s result over to the whole chain

    What it costs: Southern and northern stores buy different sizes and seasons

    How to avoid: Check the curve by store cluster.

  • A curve by sales at incomplete availability

    What it looks like (illustrative figures): Scarce sizes are understated.

Russia and the EAEU

In Russia moving stock between a chain’s stores regularly runs into distance and time: goods travel to a store thousands of kilometers away for weeks, and by the time they arrive the seasonal demand for that size has passed. That is why a move is compared with a markdown on the spot, and for remote stores it often pays better to pass stock to online sales, where a size looks for a buyer across the whole country.

Transfers are documented with primary papers: within one legal entity it is an internal transfer, between different legal entities it is a sale with 22% VAT and a universal transfer document (UPD) in electronic document exchange. For marked goods (footwear, apparel, light industry) the transfer must be reflected in the Chestny ZNAK marking system with the codes, otherwise stock in the books and in the marking system diverges; check the procedure with your accountant. On marketplaces a size is searched through the seller’s size chart: make sure the mapping of sizes (RU, EU, US, UK, CN) is right, otherwise “size did not fit” returns will eat the gain from evening out stock. Children’s sizes are counted by height, not by girth: check the table of the specific brand.

Size charts in the union countries are mostly built on Russian and European systems, but in Kazakhstan, Kyrgyzstan and Armenia Turkish and Chinese charts also appear on shelves: check the conversion table before moving or replenishing. Moving goods between EAEU countries is not an internal transfer but a supply with indirect taxes under Annex 18 to the EAEU Treaty (VAT in the country of destination, an import statement); a guideline, check with your accountant. Marking and VAT rates differ by country; VAT rates are on the page “EAEU and CIS countries” (a guideline as of 14.11.2026, verify).

Categories and audiences

Apparel
Apparel has 6 to 8 sizes, demand clusters around the middle of the run and the extreme sizes sell slowly. The core usually consists of three or four neighboring sizes, and skew comes from the fit of the model and from the region. Check the curve per fit: slim and relaxed models give different cores.
Footwear
In footwear the shopper is tied to the size and does not substitute a neighbor: a hole in the core costs almost the whole sale. The run is longer, some brands use half-size steps, so the core is wider and moving a pair costs more because of the box. “Size did not fit” returns are high, so moving should be counted together with the chance of return.
Accessories
Most accessories have a short size run or none (bags, scarves, jewelry), and the size topic comes down to a range question: what to remove so that sales are not diluted. Exceptions are belts, gloves and hats, which have a run. Here cannibalization by a new item matters more than run completeness.
Kids
A child’s size is set by height and the child outgrows it within a season: the core shifts during the season and leftover small sizes lose demand faster than adult ones. Demand peaks before 1 September and the New Year, so the core must be replenished before the peak. The adult buys a size “to grow into”, and the curve should be built on that behavior.
Men
For men, size is often set on two axes: shirts by collar and height, trousers by waist and length. A gap on one axis breaks the sale even when the other size is in stock: trousers at waist 32 are there, but the needed length is not. The curve is counted on both axes at once and the core is held on every basic model.
Women
For women a size is set not only by the number but by the cut: one person takes 44 in a knit dress and 46 in a fitted jacket. So the curve is built by groups of silhouettes (knit, woven, denim), not one for the whole category. Returns for “wrong size” by silhouette show whether the mistake is in the cut or in the buy.

Questions this topic answers

  • Which sizes give most of the sales and where are we empty?
  • Can we even out stock by moving it rather than by markdown?
  • Does the new item add sales or take them from neighbors?
  • Does a region need its own size curve?

Key measures and terms

Study first

What next

Related: “Replenishment and stock cover”, “Store and chain”, “Working with stock: transfers, top-up, defects, prices, order entry”.

Documents on this topic

Open interactively →