Syntha Lab RU

Ultra-fast and flexible chains: test-and-repeat, on-demand, near sourcing

The document compares the economics of a large batch and a micro-batch with repeat: when a premium for speed and small volume pays back through lower unsold risk. It describes test-and-repeat, on-demand production, near sourcing and the conditions under which a flexible chain beats a distant cheap one.

Brand, customer, organisation Brand and retail Verified: 2026-11-14

Open interactively →

Why this document

The commercial and production directors read this when they decide how much to order upfront and how much to hold back for repeats. It matters because a large batch is cheaper per unit but the whole risk of leftover stock stays with the brand, while a micro-batch with a speed premium lets you order little, watch demand and reorder what sells. So the benefit is counted not by unit price but by expected loss on unsold goods. Without the calculation, the speed premium is either paid across the whole range or refused where it would pay back.

At a glance

What is this document and why is it needed?
The document compares the economics of a large batch and a micro-batch with repeat: when a premium for speed and small volume pays back through lower unsold risk. It describes test-and-repeat, on-demand production, near sourcing and the conditions under which a flexible chain beats a distant cheap one.
Who owns the document and how often is it rebuilt?
Maintained by the production and sourcing director with the commercial and finance directors; the chain decision is made per group (basics, fashion, capsules) each season; results are shown to the owner as expected profit and worst case.
What mistakes are the most common?
Comparing only unit price -> a large batch always looks cheaper -> compare expected profit and worst case. A repeat with no lead time or right -> the factory is busy and demand lost -> a framework contract for repeats. A first order that is too small -> unreliable sales data -> a test on enough stores and weeks.

For whom

Brand
A brand chooses what share of the collection to order in a large batch from a distant factory and what in micro-batches nearby: basics in a large batch, fashion and capsule items by test and repeat. It needs factories with low minimums, a materials buffer and quick approvals.
Retail
A retailer uses flexibility in buying: a small first order, a reorder from a supplier with a short lead time, a fast flow from nearby warehouses and own private label micro-launches. Conditions: right to repeat, size guarantee in the repeat, reorder price.

What it consists of

1. Flexible chain models

  • Test-and-repeat: a small first batch, sales read over 1–2 weeks, reorder on actuals; the repeat share rises for strong styles.
  • On-demand: production after the customer order (customization, pre-order); almost no stock, but a long delivery time and high unit price.
  • Brand: Near sourcing: production close to the sales market; shorter lead time and less delivery risk, but a higher price.

2. What flexibility costs

  • A unit price premium for small volume, resetting, urgency and expensive remaining stock materials; higher management and planning cost.
  • Lost sales: while the repeat is in progress (the lead time), part of demand leaves; counted as the share of demand lost during the wait.

3. What we gain

  • Less unsold stock and markdown, quicker response to demand, lower inventory, higher full-price share; especially for fashion items with uncertain demand.
  • Demand data before large commitments; optionality: you pay for the right to decide later.

4. Comparing options across scenarios

  • Demand scenarios (flop, weak, plan, hit) with probabilities; profit of each option in each scenario; expected profit and worst result.
  • Criteria: maximum expected profit at an acceptable worst result; basics favor a large batch, uncertain demand favors micro-batches.

5. Operating conditions and contracts

  • Brand: Factories with low minimums and short cycles; framework contracts for repeats; stock of raw material and shared materials across styles; quick sample approval.
  • Data and process: daily sales reading, a repeat decision rule, a predefined repeat budget.

6. Brand and retailer: who pays for flexibility

  • Brand: A brand flexible chain: factories with small minimums, a framework repeat agreement, a stock of shared raw material for several styles; the brand pays the small-lot premium and answers for the repeat lead time to clients.
  • Retailer: Retailer flexibility is more often achieved through terms with partner brands: reorder, return, consignment, delivery from the brand warehouse within days; for a private label the retailer acts as a brand.
  • Brand: Brand near-sourcing: compare landed cost with duty and logistics, not the factory price; the gain in lead time is converted to money through lost demand and leftovers.
  • Retailer: Retailer fast replenishment: transfers between stores, a reorder from the brand warehouse; the cost of a lost sale during delivery is compared with the cost of speeding up.
  • Brand: Pre-order and made-to-order: a prepayment helps finance a lot but binds you on dates; return and cancellation terms are set in the offer.
  • Retailer: Retailer pre-order: a reservation of the window and stock, return rules for distance selling; it is important not to sell more than the supplier brand confirmed.
  • Brand: The brand repeat decision rule: a sales-speed threshold from wholesale reorders and D2C, a repeat budget per style, a 5-7 day decision window.
  • Retailer: The retailer decision rule: a week-two sell-through threshold for a reorder, redistribution across stores and online before reordering, a reorder cap from OTB.

How to read the result and which decisions follow

  • For women expected profit is RUB 4.72 m for the large batch and RUB 4.26 m for the micro-batch, and in a flop 0.30 against 1.14. Decision: the micro-batch is costlier on average but protects against a flop; the choice depends on what is more dangerous for you, the average loss or the worst outcome.
  • Brand: the allowed repeat price is RUB 1,468 against a large-batch unit cost of RUB 1,350, a premium for speed of 9%. Decision: agree with the factory a repeat no costlier than this threshold and guaranteed slots; otherwise the micro-batch loses its point.
  • Retailer: an ultra-fast chain is available to you through a test in part of the stores and a quick top-up from brands on the base; decide on buying shallower depth only where the supplier guarantees the repeat term.
  • Brand: in men and accessories the large batch wins even in the worst outcome: leftovers sell better there (men classics, accessories without sizes). Decision: apply ultra-fast to fashion items with a high flop risk.
  • Retailer: the sensitivity table for women shows the difference from −0.24 to −RUB 0.75 m: the more demand is lost waiting for the repeat and the costlier the repeat, the better the large batch. Measure lost demand in a test.

Formula derivation and sensitivity tables

  • Derivation of the allowed repeat price. The micro-batch expected profit E_M(c_R) = Σ q_k x [(Q_1 + R_k) x p - Q_1 x c_T - R_k x c_R] is linear in c_R with a slope of -Σ q_k x R_k. Equating it to E_L gives c_R* = (A - E_L) / Σ q_k x R_k, where A is E_M at c_R = 0.
  • Derivation of the worst outcome: in a flop scenario D < Q the large batch loses Q x c_L - D x p - (Q - D) x s, while the micro-batch only Q_1 x c_T - D x p - (Q_1 - D) x s at D <= Q_1; the difference is set by batch size, so the micro-batch protects roughly in proportion to Q_1 / Q.
  • Numbers (women): expected profit of the large batch RUB 4.72 m, of the micro-batch RUB 4.26 m; the allowed repeat price is RUB 1,468 (root found by code, 60 bisection iterations).
  • Numeric table of micro minus large, RUB m (rows lost demand 6..18%, columns repeat price -20..+20%): 6%: 0.21 / −0.03 / −0.27 / −0.51 / −0.75; 9%: 0.10 / −0.13 / −0.37 / −0.60 / −0.83; 12%: −0.01 / −0.24 / −0.46 / −0.69 / −0.91; 15%: −0.13 / −0.34 / −0.56 / −0.78 / −1.00; 18%: −0.24 / −0.45 / −0.66 / −0.87 / −1.08.
  • Brand: each three percentage points of lost demand shift the micro-batch gain down by RUB 0.10 m (between the rows of 12% and 15%); so the demand lost while waiting for the repeat must be measured, not taken as zero.

Control checks

  • Check: the scenario probabilities sum to 1.00, expected profit of 4.72 and RUB 4.26 m matches the table (code).
  • Brand: demand scenario probabilities are assumptions; revise them from the sales of test batches of past seasons.
  • Retailer: the leftover price for sale in the large batch must be realistic (RUB 900 at a price of RUB 4,200 for women); an overstated leftover price hides risk.
  • Brand: nearshoring and on-demand change the unit price and lead time; compare them by the same criterion (expected profit and worst outcome).
  • Retailer: set the selection rule (by expected profit or worst outcome) in advance; choosing during the calculation fits the result.

Where the data comes from

Data sourceFields
First days and weeks of salesSales by style and size, stock, sales velocity
Factory termsBatch and repeat price, minimums, lead time, material stock
Markdown and leftover historyShare unsold and leftover price by styles of past seasons
Collection calendarSelling window, repeat time, last decision date

How to build it step by step

  1. Step 1. Split the assortment into basics, fashion and capsules by demand predictability.
  2. Step 2. For each group collect demand scenarios and probabilities from history.
  3. Step 3. Collect prices and lead times: a large batch from a distant factory and a micro-batch with repeat nearby.
  4. Step 4. Determine the leftover price and the share of demand lost during the repeat.
  5. Step 5. Compute each option profit by scenario, expected profit and worst result.
  6. Step 6. Find the allowable premium on the repeat price at which the options are equal.
  7. Step 7. Choose an option per group; for fashion reserve a share for repeats.
  8. Brand: Step 8. Agree with factories on repeat right, time and price; record this in a framework contract.
  9. Step 9. Set a repeat decision rule: a sales velocity threshold and decision date.
  10. Step 10. After the season compare plan with actuals: leftovers, losses, flexibility premium paid.
  11. Brand: Step 11. Brand: for each group compute the repeat price at which the micro-batch equals the large batch and write it into the framework agreement as an upper limit.
  12. Retailer: Step 12. Retailer: apply the repeat rule by sales speed to fashion items and compare the result with buying one large lot for the season.

Formulas

MeasureFormulasNote
Brand: Large batch profitProfit = min(Q; D) x price + max(Q - D; 0) x salvage price - Q x unit costQ is the batch, D demand; salvage price may be below cost.
Brand: Micro-batch with repeat profitRepeat R = (D - Q1) x (1 - lost demand share); profit = (Q1 + R) x price - Q1 x test cost - R x repeat costIf D is below Q1, the leftover of the first batch is sold at the salvage price.
Brand: Expected profitE = sum p_i x profit_iProbabilities sum to 100%; estimated from history and revised each season.
Brand: Allowable repeat premiumFind the repeat price at which micro-batch E equals large batch EIf the factory asks for more, use the large batch or look elsewhere.

Formulas in notation

Large batch profit
Π_L = min(Q, D) × p + max(Q − D, 0) × s − Q × c_L
Q is the batch, D demand, p price, s the leftover price, c_L cost in the large batch.
Micro-batch profit with repeat
Π_M = (Q_1 + R) × p − Q_1 × c_T − R × c_R, R = (D − Q_1) × (1 − λ)
Q_1 is the test batch, λ the share of demand lost while waiting for the repeat, c_T and c_R the test and repeat costs.
Expected profit
E[Π] = Σ q_k × Π(D_k)
q_k are the probabilities of demand scenarios; they sum to 1.
Allowed repeat price
c_R*: E[Π_M](c_R*) = E[Π_L]
The repeat cost at which the micro-batch and the large batch are equal in expected profit.

Worked example

Exercise: large batch or micro-batch with a repeat by category, RUB m

CategoryPrice, RUBLarge batch, pcsMicro-batch, pcsExpected profit, largeExpected profit, microWorst case, largeWorst case, microChoice
Women4,2003,0006004.724.260.301.14Micro-batch on the worst case
Men4,8002,0005006.154.884.802.76Large batch
Kids2,5004,0001,2004.894.291.682.17Micro-batch on the worst case
Footwear6,5001,5004003.442.900.961.20Large batch
Accessories3,0002,5008004.723.753.811.74Large batch
Total23.9320.09

Notional: each category has its own price, a four-scenario demand spread with probabilities, a salvage price and a share of demand lost while the repeat runs (women equals the base example: price RUB 4,200, 3,000 pcs against 600 + a repeat). Choice rule: a micro-batch if its expected profit is higher; a micro-batch on the worst case if it loses less on a flop and trails on expectation by no more than 15%; otherwise the large batch. Summed expected profit is RUB 23.93m with a large batch everywhere and RUB 20.09m with a micro-batch everywhere; it is better to choose by category. Menswear basics keep their price when unsold, so paying a flexibility premium makes little sense there; in footwear the small-lot premium is high because of the last. All inputs are exercise assumptions.

Figures

Expected profit: large batch and micro-batch
Bar chart: Women 4.72, Men 6.15, Kids 4.89, Footwear 3.44, Accessories 4.7202468WomenWomenLarge batch: 4.724.72Micro-batch with repeat: 4.264.26MenMenLarge batch: 6.156.15Micro-batch with repeat: 4.884.88KidsKidsLarge batch: 4.894.89Micro-batch with repeat: 4.294.29FootwearFootwearLarge batch: 3.443.44Micro-batch with repeat: 2.92.9AccessoriesAccessoriesLarge batch: 4.724.72Micro-batch with repeat: 3.753.75RUB m
  • Large batch
  • Micro-batch with repeat
Data as a table
MeasureLarge batchMicro-batch with repeat
Women4.724.26
Men6.154.88
Kids4.894.29
Footwear3.442.9
Accessories4.723.75

Expected profit: large batch and micro-batch. The average over demand scenarios in RUB m: the large batch wins in three categories of five.

Worst outcome: large batch and micro-batch
Bar chart: Women 0.3, Men 4.8, Kids 1.68, Footwear 0.96, Accessories 3.81012345WomenWomenLarge batch: 0.30.3Micro-batch: 1.141.14MenMenLarge batch: 4.84.8Micro-batch: 2.762.76KidsKidsLarge batch: 1.681.68Micro-batch: 2.172.17FootwearFootwearLarge batch: 0.960.96Micro-batch: 1.21.2AccessoriesAccessoriesLarge batch: 3.813.81Micro-batch: 1.741.74RUB m
  • Large batch
  • Micro-batch
Data as a table
MeasureLarge batchMicro-batch
Women0.31.14
Men4.82.76
Kids1.682.17
Footwear0.961.2
Accessories3.811.74

Worst outcome: large batch and micro-batch. Profit in the flop scenario in RUB m: the micro-batch protects in women and kids, in men and accessories the opposite.

Women: profit by repeat cost
Line chart: Micro-batch: 5.29 → 1.65; Large batch: 4.72 → 4.7201234561,0001,5002,0002,5003,0003,5004,000threshold about RUB 1,468Micro-batch, 1000: 5.29Micro-batch, 1500: 4.68Micro-batch, 2000: 4.08Micro-batch, 2500: 3.47Micro-batch, 3000: 2.86Micro-batch, 3500: 2.25Micro-batch, 4000: 1.65Large batch, 1000: 4.72Large batch, 1500: 4.72Large batch, 2000: 4.72Large batch, 2500: 4.72Large batch, 3000: 4.72Large batch, 3500: 4.72Large batch, 4000: 4.72RUB m
  • Micro-batch
  • Large batch
Data as a table
MeasureMicro-batchLarge batch
1,0005.294.72
1,5004.684.72
2,0004.084.72
2,5003.474.72
3,0002.864.72
3,5002.254.72
4,0001.654.72

Women: profit by repeat cost. The micro-batch beats the large batch as long as the repeat costs below the threshold; the crossing gives the allowed premium for speed. Horizontal axis: unit cost of the repeat, RUB.

Women: micro-batch minus large batch as lost demand and repeat cost vary
Heat map: 5×5; min −1.08, max 0.21Repeat -20%Repeat-20%Repeat -10%Repeat-10%Repeat 0%Repeat0%Repeat +10%Repeat+10%Repeat +20%Repeat+20%Lost demand 6%Lost demand 6%0.21Lost demand 6% / Repeat -20%: 0.21−0.03Lost demand 6% / Repeat -10%: −0.03−0.27Lost demand 6% / Repeat 0%: −0.27−0.51Lost demand 6% / Repeat +10%: −0.51−0.75Lost demand 6% / Repeat +20%: −0.75Lost demand 9%Lost demand 9%0.1Lost demand 9% / Repeat -20%: 0.1−0.13Lost demand 9% / Repeat -10%: −0.13−0.37Lost demand 9% / Repeat 0%: −0.37−0.6Lost demand 9% / Repeat +10%: −0.6−0.83Lost demand 9% / Repeat +20%: −0.83Lost demand 12%Lost demand 12%−0.01Lost demand 12% / Repeat -20%: −0.01−0.24Lost demand 12% / Repeat -10%: −0.24−0.46Lost demand 12% / Repeat 0%: −0.46−0.69Lost demand 12% / Repeat +10%: −0.69−0.91Lost demand 12% / Repeat +20%: −0.91Lost demand 15%Lost demand 15%−0.13Lost demand 15% / Repeat -20%: −0.13−0.34Lost demand 15% / Repeat -10%: −0.34−0.56Lost demand 15% / Repeat 0%: −0.56−0.78Lost demand 15% / Repeat +10%: −0.78−1Lost demand 15% / Repeat +20%: −1Lost demand 18%Lost demand 18%−0.24Lost demand 18% / Repeat -20%: −0.24−0.45Lost demand 18% / Repeat -10%: −0.45−0.66Lost demand 18% / Repeat 0%: −0.66−0.87Lost demand 18% / Repeat +10%: −0.87−1.08Lost demand 18% / Repeat +20%: −1.08
  • −1.08 … 0 … 0.21 RUB m
Data as a table
Repeat -20%Repeat -10%Repeat 0%Repeat +10%Repeat +20%
Lost demand 6%0.21−0.03−0.27−0.51−0.75
Lost demand 9%0.1−0.13−0.37−0.6−0.83
Lost demand 12%−0.01−0.24−0.46−0.69−0.91
Lost demand 15%−0.13−0.34−0.56−0.78−1
Lost demand 18%−0.24−0.45−0.66−0.87−1.08

Women: micro-batch minus large batch as lost demand and repeat cost vary. The difference of expected profit in RUB m: negative values mean the large batch is better; rows the lost demand share, columns the repeat price ±10% and ±20%.

Specifics by category

Apparel
The main candidate for test-and-repeat in fashion: a small first batch (in the example 600 of 3,000 pcs), reading sales for 1-2 weeks, a reorder on actuals. The small-lot premium of about 30% over cost (1,750 against RUB 1,350, an assumption) pays back through lower leftovers and a higher full-price share. Basics (T-shirts, knitwear) are more efficient in a large batch; for fashion and capsules a micro-batch narrows the worst case. Watch the repeat lead time: demand lost while waiting is 12% in the example.
Footwear
Flexibility is limited by tooling: lasts and molds pay back through the run, so a micro-batch pair is clearly costlier (an assumption: 3,500 against RUB 3,000) and the size run splits a small batch into 11 sizes of 35-40 pairs. A repeat is realistic for styles with an existing last; for a new style near-sourcing does not shorten last making. Leftovers in slow sizes sell worse than apparel, so the salvage price is lower (an assumption of 40% of price).
Accessories
There are no sizes, so a small series does not break on the size grid: colors and materials are tested in small series. Leftovers of basic accessories keep their price (an assumption of 70% of price), the flop risk is small and the large batch often wins on expectation. On-demand suits engraving, initials and gift sets. Gift demand before December and the spring holidays is rigid: a repeat must land a week before the date, otherwise it is useless.
Kids
Demand is tied to windows (school, New Year), so the share of demand lost while the repeat runs is higher: 15% against 12% in adults (an assumption); a repeat after 1 September is pointless. The 6 cm height run spreads a small batch over 3-4 sizes, and unsold sizes do not carry over to next year. A micro-batch is justified for fashion and gifts; school uniform and basics go in a large batch delivered by July. Conformity documents are needed for each new lot and style.
Men
Demand for basics is stable: the scenario spread is narrow (in the example 1,000-2,300 pcs around a plan of 1,900), so the large batch wins on expectation and basics leftovers keep their price (an assumption of 63% of price). There is no point paying a flexibility premium; a framework repeat agreement and auto-replenishment are more useful. Flexibility is for seasonal novelties and capsules; for suits and shirts near-sourcing may be justified by short delivery, not by a micro-batch.
Women
The main application zone: a wide demand spread (in the example a flop of 500 against a hit of 3,600 pcs). A micro-batch strongly limits the flop loss and the repeat catches the hit, but in the example the large batch wins on expectation: flexibility costs a premium and lost demand. The choice depends on the acceptable worst case and the flop share. The allowable repeat price at which expected profits are equal is computed in the example result. Seasonality and trend shorten the repeat window to a few weeks.

Common mistakes

  • Comparing only unit price -> a large batch always looks cheaper -> compare expected profit and worst case.
  • A repeat with no lead time or right -> the factory is busy and demand lost -> a framework contract for repeats.
  • A first order that is too small -> unreliable sales data -> a test on enough stores and weeks.
  • Lost demand during the repeat ignored -> profit overstated -> include the lost demand share in the model.
  • Flexibility for the whole collection -> overpaying for basics -> a large batch for predictable demand.

Russian specifics

  • Near sourcing in EAEU countries falls under common technical regulations (TR CU 017/2011) and union customs rules; imports from the EAEU are free of duty but follow indirect tax rules. Rule: CU Commission decision 876 of 09.12.2011: TR CU 017/2011 on safety of light-industry products (Customs Union Commission Decision of 09.12.2011 No. 876) (rule check not completed).
  • Each new batch of light industry goods receives marking codes: micro-batches and repeats increase the number of code orders and operations; include this in operating costs. Rule: Decree 1956 of 31.12.2019: Government Decree of 31.12.2019 No. 1956 (marking rules for light-industry goods).
  • Pre-order and made-to-order terms (prepayment, term, return) must comply with sale rules and the consumer protection law; individually made goods have special return rules, check against the current wording. Rule: Law 2300-1 of 07.02.1992: Law of 07.02.1992 No. 2300-1 on consumer rights protection (rule check not completed).
  • Brand: each micro-batch and repeat gets its own marking codes and, for kids, a conformity confirmation; the number of documents grows with the number of lots, so include their cost in the flexibility premium (the procedure is unverified).
  • Retailer: for distance sale on pre-order the return and buyer-information rules apply; the delivery date is stated in the offer and a breach lets the buyer cancel (details unverified).

Open interactively →