Syntha Lab RU

Collection lifecycle and the critical path

Eight stages from plan to season close: who owns each, which system owns the data, how long the stages really take and what share of orders suppliers deliver on time and in full.

both Processes and data

Open interactively →

About the topic

A collection lives along one route, and nearly all season failures happen not at the thinking stage but at the joints between stages. This topic shows the whole route: plan, design and tech pack, materials purchase and order, production, import and labeling, warehouse and distribution, sales, season close. Each stage needs three things: an owner who answers for the result, a system where the stage data lives, and a real duration, not the one from the presentation.

Why it matters to a brand or retailer. A delivery two weeks late looks harmless in a spreadsheet, but it means seasonal goods arrive when weather and demand have already moved, and part of the collection goes to markdown. The decisions here are: at which stage to build a time buffer, where to put control, who owns the data at a joint, which suppliers get rush orders and whether to change a factory that is consistently late.

The reasoning rests on three ideas. First: the collection path is a chain, and its length is set by the longest sequence of stages (the critical path), not by their sum and not by an average. Speeding up a stage that is off the critical path gives nothing. Second: duration is not a number but a spread. A factory that on average takes 45 days but arrives 10 days late in a third of orders breaks the schedule more than a factory with a steady 50 days. So you look not only at the mean but at the worst quartile. Third: delivery performance has two conditions at once: on time and in full (OTIF). An order that came on time but with 70% of the pieces and an order that came complete but a week late both count as failed; otherwise a supplier looks good on either measure alone.

Where to start. Draw your stages, assign an owner to each, list the system that owns the data, take last season and compare planned and actual dates. It takes a week and immediately shows where durations truly diverge from plan. Then compute OTIF by supplier and place a time buffer exactly at the stages where the divergence is systematic.

Traps: counting duration by the average supplier, failing to agree with the supplier what on time and in full mean, leaving a stage without an owner, planning from the start date instead of the shelf date, and treating a stage as closed without a document or event in the system. A separate trap for imports: customs and labeling are often missing from the factory schedule, though they are weeks of real time.

Key ideas

An owner and a system of record for every stage

If a stage has no single owner, data is lost at the joints. Illustratively: three hand-offs (tech pack → order → warehouse), each passing 95% of records without error; without rework 0.95 × 0.95 × 0.95 ≈ 85.7% reach the warehouse, so every seventh record needs fixing. Name an owner and a system of record for every hand-off.

Duration as the sum of stages plus overrun

Plan: tech pack 25 days, materials 30, sewing 45, transit and customs 25, warehouse and labeling 15, total 140. Last season actual: 30, 36, 50, 32 and 18, total 166, overrun 26 days. A 7-day buffer covers only part, lateness is 19 days. Plan from the actual, not from the plan.

Critical path: speed up what sets the duration

Fabric takes 30 days and cutting 5, so the chain is 35 days, while trims are ordered in 20. Speeding trims by 10 days does not change the duration: the path is still 35. Cutting fabric by 5 days moves the whole date by 5. Always ask which chain is the longest and with what slack the parallel one runs.

OTIF counts on time and in full

Ten supplier orders: 8 arrived on time, 9 in full, and both held for 7. Perfect orders are 70%, not 80 or 90%. If the events were independent you would expect 0.8 × 0.9 = 72%. Agree the definitions of on time (a tolerance in days) and in full (a tolerance in pieces) with the supplier in advance and fix them in the contract.

The price of lateness is counted in season margin

Season sales of 30 million roubles excl. VAT over 120 days give 250,000 roubles per day; at 55% margin that is 137,500 roubles of margin per day. A delay of 19 days under the worst assumption that sales cannot be recovered costs about 2.6 million roubles of margin. The real loss is smaller if part of demand shifts, but the order of magnitude shows what a time buffer or rush air freight is worth.

How to work through it

  1. Draw your collection’s eight stages and mark the joints where data passes from system to system.
  2. Assign an owner to every stage and a system of record for its data. One stage, one person accountable.
  3. Take last season and compare planned and actual dates by stage; find the stage with the biggest overrun.
  4. Find the critical path and the parallel chains: what actually sets the shelf date.
  5. Compute OTIF by supplier after defining on time and in full with them, and single out the worst one.
  6. Place the time buffer at stages with systematic overrun, not evenly on all of them.
  7. Estimate the price of lateness in margin and compare it with the cost of speeding up; decide where to pay for urgency.
  8. Fix checkpoints: a stage counts as closed only by a document or an event in the system.

Where to start

1) Draw your own stages on the “End-to-end collection process” scheme. 2) Assign an owner to each stage. 3) Name the system of record for the stage data. 4) Measure the real stage durations of the last season. 5) Calculate supplier OTIF. 6) Find the stage with the biggest lead-time overrun. 7) Fix a time buffer on that stage. Calculations of the topic in order: “Supplier OTIF and the lead-time stack”. See also (their primary topic is another one): “Collection calendar: critical path and slack”, “Collection gates (stage-gate): a funnel of hypotheses, the price of a hit, the saving against “everything into a trial” and the stop rule of the pilot batch”.

Calculations in this topic

  • Supplier OTIF and the lead-time stack

    From supplier orders calculates the share on time, in full and both (OTIF), perfect orders, the worst supplier, average lead time planned and actual and by stage, and which stage used up the time buffer.

    What share of orders arrives on time and in full and which stage uses up the time buffer.

  • Collection calendar: critical path and slack

    Finds the longest chain of stages from idea to warehouse, shows which stages have float and compares the path with the buffer and the date goods are needed in the warehouse. Negative slack says how many days must be squeezed.

    Which stages move the launch date, how much slack there really is and how many days must be squeezed.

  • Collection gates (stage-gate): a funnel of hypotheses, the price of a hit, the saving against “everything into a trial” and the stop rule of the pilot batch

    From the number of ideas, the cost of the stages per model and the share of models passing each gate (idea, tech pack, sample, pilot batch) it counts how many models reach a hit, how much is spent in all, what one hit costs, what part of the spend went to rejected models and how much the gates save against “all ideas go to the end”. Separately it counts the stop rule of the pilot batch: the sales threshold at the control week, the loss on stopping and the loss avoided by not scaling blindly. Take the pass shares from past seasons; here illustrative.

    How many models will reach a hit, what one hit costs, how much the gates save and at what pilot result the batch is stopped.

Self-check questions

  • Is every one of the eight stages assigned an owner, and does the owner know it?
  • In which system does each record live at each stage and does the SKU key match?
  • Which stage overran its planned duration the most last season and by how many days?
  • What is the critical path length and which stages run in parallel with slack?
  • How did you and the supplier define on time and in full?
  • Does the time buffer cover the typical overrun, not the supplier average?
  • Are customs, labeling and holidays of the factory country included in the schedule?
  • What data is handed over at the joints and who answers for its completeness?
  • How much margin does a day of lateness cost and how much does it cost to speed delivery up?

Knowledge check

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

  1. The lead-time stack plan by stage: materials 20 days, production 30, transit and customs 25, receiving 3; actual: 26, 33, 27 and 4. The time buffer is 7 days. How many days of overrun are not covered by the buffer and which stage exceeded its plan the most?

    • A. 12 days; materials the most
    • B. 5 days; materials the most
    • C. 5 days; production the most
    • D. 6 days; materials the most

    Correct answer: B. 5 days; materials the most. The plan is 20 + 30 + 25 + 3 = 78 days, the actual 26 + 33 + 27 + 4 = 90, the overrun is 12 days; not covered is 12 − 7 = 5 days. Materials exceeded the plan the most (+6 days against +3, +2 and +1). 12 days is the overrun without deducting the buffer; production is the longest stage in the plan but not the biggest overrun; 6 days is the overrun of one stage, not the total of the chain.

  2. Supplier A: 50 orders, 45 on time, 48 in full, 44 on time and in full. Supplier B: 30 orders, 21 on time, 27 in full, 20 on time and in full. What is the overall OTIF and who is weaker?

    • A. 82.5%; B is weaker
    • B. 77.3%; B is weaker
    • C. 80%; B is weaker (66.7%)
    • D. 93.8%; B is weaker

    Correct answer: C. 80%; B is weaker (66.7%). OTIF = (44 + 20) / (50 + 30) = 64 / 80 = 80%; by supplier A has 88% and B 66.7%, B is weaker. 82.5% is the on-time share ((45 + 21) / 80), 93.8% is the in-full share ((48 + 27) / 80), and 77.3% comes from averaging 88% and 66.7% without the number of orders.

  3. 20 ideas, the gate pass shares 60, 50, 60 and 35%. How many models become hits?

    • A. 4.2
    • B. 1.26
    • C. 7.0
    • D. 0.35

    Correct answer: B. 1.26. The shares multiply: 0.6 × 0.5 × 0.6 × 0.35 = 0.063, and 20 × 0.063 = 1.26. 4.2 comes if only the last gate is taken (35% of 12), and 7.0 is just 35% of 20 ideas.

  4. The funnel spent RUB 1,637,000 and gave 1.26 hits. What does one hit cost?

    • A. 257,500 ₽
    • B. 1,637,000 ₽
    • C. 1,299,206 ₽
    • D. 81,850 ₽

    Correct answer: C. 1,299,206 ₽. 1,637,000 ÷ 1.26 = RUB 1,299,206: all the funnel’s spending, including the rejected models, is divided by hits. RUB 257,500 is the cost of one model’s pilot, and RUB 1,637,000 is the spending without dividing.

  5. A demand test on a batch of 150 units, a sell-through threshold of 20% by week 3. How many units must be sold by the control week?

    • A. 20 units
    • B. 30 units
    • C. 75 units
    • D. 120 units

    Correct answer: B. 30 units. 150 × 20% = 30 units. The threshold is counted from the pilot batch and not from the scale: 75 is 50%, 120 is 80% of the batch, and 20 units takes the percentage without multiplying by the batch.

Common mistakes

  • Mistake: Counting duration by the average supplier

    What it costs: The worst suppliers break the schedule unexpectedly

    How to avoid: Look at the spread and the worst quartile.

  • Mistake: Failing to agree on time and in full

    What it costs: You and the supplier count differently and argue instead of deciding

    How to avoid: Fix tolerances in days and pieces in the contract.

  • Mistake: Leaving a stage without an owner

    What it costs: Data is lost at the joint and fixes happen in chats

    How to avoid: One accountable person and one system of record.

  • Mistake: Planning from the start date, not from the shelf date

    What it costs: Only convenient durations are added

    How to avoid: Count backwards from the shelf date.

  • Mistake: Treating a stage as complete without a document or system event

    What it costs: Done on paper surfaces at the warehouse

    How to avoid: Require confirmation in the system.

  • Mistake: Moving the stop threshold after sales start

    What it costs: A weak model goes to scale On a failure the scale costs RUB 1.31 million against RUB 87,000 when stopping at the threshold.

    How to avoid: Record the threshold in units before the first sale The head writes the threshold before sales start and checks against it without debate.

    What it looks like (illustrative figures): The threshold is 30 units by the control week; 24 are sold, the threshold is “lowered to 20” and a scale of 3,000 units is ordered.

Russia and the EAEU

In Russia, the production cycle gains stages that are not in the factory schedule. Import needs customs clearance, and import VAT (standard rate 22% from 01.01.2026) must be financed before release. For product groups under mandatory labeling (the Chestny ZNAK system, Government Decree No. 1956) codes are put into circulation before sale, and a batch cannot go to a store without codes; the procedure, waves and transition periods for remaining stock have changed, so check dates against the current text on the collection launch date (checked 14.11.2026). Mandatory conformity assessment under the Customs Union technical regulations (TR CU 017/2011 for light industry, TR CU 007/2011 for children’s goods) must be built into the time before shipment.

The calendar of the factory country and your own also shift schedules: Chinese New Year usually halts production for several weeks, while May and New Year holidays in Russia shorten working days at receiving. Documents between participants travel through electronic document exchange (UPD, UKD), and an exchange failure delays acceptance as much as a supply failure. Put these steps into the schedule as separate lines, and take broker and warehouse durations from your own experience, not from advertising promises.

Inside the EAEU there are no duties, but import stages do not vanish: you need an import declaration and indirect-tax documents (VAT under destination-country rules). Labeling runs on each country’s own schedule. Reference for Kazakhstan as of 14.11.2026, confirm the order’s text: rules in force from 14.11.2026, mandatory labeling of the first stage from 01.12.2026, the second from 01.03.2027, the third from 01.10.2027, remaining stock until 01.01.2028. For Belarus, Armenia and Kyrgyzstan national dates: indicative as of 14.11.2026, check the text of the act, so for deliveries to Russia ask who orders the codes: the importer in Russia or the sender. Put 1–3 weeks into the cycle for documents and codes and test them on an actual batch.

Categories and audiences

Apparel
Apparel passes all eight stages and most often breaks at fabric and sewing: fabric has a long cycle and minimum lots, and a late color decision shifts everything. Keep buffer at material purchase and monitor cutting in parallel with trims.
Footwear
Footwear has a long path for lasts and tooling: a new last and patterns add weeks before the sample, and the season is tightly tied to weather. A shift at the sample stage is hard to recover, so buffer goes there first. The size run affects warehouse and distribution.
Accessories
Accessories often have a shorter cycle, and the key risk is not time but sourcing components and hardware the factory may lack. Light weight allows rush air delivery without hurting cost much, so the price of speeding up is lower.
Kids
Children’s seasons are rigid: school in late summer and New Year; a week’s delay at once becomes a missed peak. Add a certification stage for children’s goods requirements and labeling: they lengthen the path. Set checkpoints earlier than for adults.
Men
In the men’s line it is worth running two routes: carried-over basics follow a proven chain, “repeat the last order”, while novelties go through the full cycle. The risk arises when a factory is changed: for a basic style the color and fit may drift from the reference and the sample approval has to be repeated. Build a time buffer for such a case.
Women
In the women’s collection more styles are created from scratch and each goes through samples and fittings: a fit error sends the style back to the design stage. So control is placed at the junction “sample — fabric purchase”, and for future hits a short reorder is agreed with the factory in advance, so that a successful color can be repeated without a new cycle.

Questions this topic answers

  • Who owns each of the eight stages and which system owns its data?
  • Which stage exceeds its planned duration the most and by how many days?
  • What share of supplier orders arrives on time and in full?
  • Is the time buffer enough to cover the overrun?
  • Which data passes between stages and who is accountable for its completeness?

Key measures and terms

Study first

What next

Next: “The planning calendar”, “Master data: SKU, SKU, barcode”, “Collection calendar and materials”.

Documents on this topic

Open interactively →