Warehouse, fulfillment and the last mile
Own warehouse or 3PL, warehouse flows and peaks by sale stage, delivery cost by type, split shipments and delivery time against conversion.
About the topic
A warehouse and delivery turn a sale into a delivered order. The buyer sees the price of the item while the brand also pays for receiving, storage, picking, packing, delivery and return. These costs are spread across different contracts and reports, so they are easy to underestimate: the price per order in a fulfillment operator’s table looks modest until you add paid storage, stock insurance, fines for receiving errors and returns.
Why a brand or retailer needs this. Four decisions are made here: keep your own warehouse or hand goods to an operator (3PL, a third-party logistics operator that stores and ships for a fee), whether people and hours suffice for the sale peak, which delivery type to choose and whether to promise the buyer a fast delivery time. For a store with an online channel it is also the question of where to ship an order from.
How to reason. A warehouse is a set of fixed costs (rent, equipment, a manager) and variable ones (labor per order, packaging). Your own warehouse wins at high volume because the fixed costs are spread over more orders; an operator wins at low volume because you pay only for what you use. The crossing point is called the break-even volume. In an illustrative example: own warehouse RUB 600,000 a month plus RUB 90 per order, operator RUB 160 per order; the difference is RUB 70 per order, so 600,000 / 70 = 8,571 orders a month. Below that the operator is cheaper, above it your own warehouse. But the price of one month decides nothing: seasonality moves you one way and the other.
Then look at the peak. Staff are counted on the busiest period, not on the average month. A sale, a pre-holiday week and the wave of returns after it need hours that an ordinary month does not. It helps to split the periods by stage: BS before the sale, IS in the sale, AS after it, SO the final clearance (your company may use different names).
Where to start. Collect three numbers: how many orders in an ordinary month and at the peak, what an order costs you and the operator including storage and returns, and the share of non-redemption by delivery type. Typical traps: comparing tariffs without storage and returns, counting on the average month, choosing delivery by tariff rather than by full cost with non-redemption, not reading claim terms and fines in the operator’s contract.
Key ideas
Fixed and variable costs
Own warehouse: RUB 600,000 a month fixed and RUB 90 per order. Operator: RUB 160 per order with no fixed part. Break-even 600,000 / (160 − 90) = 8,571 orders a month. At 5,000 orders your own warehouse costs RUB 1,050,000 (RUB 210 per order), the operator RUB 800,000. At 12,000 orders your own costs RUB 1,680,000, the operator RUB 1,920,000.
The peak sets the headcount
If an ordinary day picks 300 orders and a sale peak picks 1,200, headcount by the average gives a fourfold shortage of people. Hours per order are multiplied by peak volume and divided by one person’s working hours with a reserve for leave and sickness. Returns after the sale (stage AS) are counted separately: receiving and repacking them uses the same hands.
Full delivery cost with non-redemption
Courier: delivery RUB 300, return RUB 200, redemption 80%. Expected cost per order 300 + 0.2 × 200 = RUB 340, per redeemed order 340 / 0.8 = RUB 425. Pickup point: delivery RUB 180, return RUB 180, redemption 65%: 180 + 0.35 × 180 = RUB 243 per order, RUB 374 per redeemed order. The courier tariff is higher but the gap after non-redemption is smaller than it looks.
Split orders
An order of three items picked at two warehouses leaves in two parcels: you pay delivery twice and the buyer gets goods on different days. If such orders are 15% and the second parcel costs RUB 250, then on 10,000 orders that is RUB 375,000 a month. A stock placement rule pays back precisely here.
A fast promise must pay for itself
Suppose a site has 100,000 visits, conversion 2.0% → 2.2% with next-day delivery, basket RUB 5,000 and margin 40%. 200 orders are added, 200 × 5,000 × 0.4 = RUB 400,000 of margin. If fast delivery costs an extra RUB 150 on 2,200 orders (RUB 330,000), the promise yields +RUB 70,000. At 2.1% conversion it already loses.
A warehouse is counted for the peak, a return by its outcomes
A ×1.6 peak gives 3,529 places and 6,353 m², low-season utilization 53%. The loss per return is RUB 539; RFID pays back in 8.8 months; an ABC layout shortens the path by 32.5%.
Formula: Places = average stock × peak coefficient ÷ units per place ÷ utilization
What to do: Peak utilization not above 90%.
How to work through it
- Collect volumes: orders in an ordinary month and at the peak, by stages BS, IS, AS, SO, and the return share.
- Add up the operator’s full cost per order: handling, storage, stock insurance, packaging, returns, fines for errors.
- Add the fixed and variable costs of your own warehouse and find the break-even volume.
- Compare with the peak: are staff and space enough in the busiest week and after it.
- Compare delivery types by full cost per redeemed order, not by tariff.
- Check split orders: share, cost of the second parcel, the option to pick in one place.
- Assess the promised delivery time: conversion lift against its price. Decide on one number, profit.
Where to start
Start with “Own warehouse or 3PL” using the operator’s offer, then “Warehouse flows” for the sale peak and “Last mile” by delivery type. Calculations of the topic in order: “Own warehouse or 3PL: cost per order and break-even volume”, “Warehouse flows: hours, staff and load by sale stage (BS, IS, AS, SO)”, “Last mile by delivery type: courier, pickup point, post”, “Split shipments: the cost of several parcels versus consolidation”, “Delivery promise and conversion: what pays for a fast promise”, “Warehouse capacity for the peak: places, area, utilization”, “RFID payback: stock accuracy, availability, stock counts”, “Returns processing: cost and restocking”, “Warehouse slotting by ABC: a shorter picking path”.
Calculations in this topic
Own warehouse or 3PL: cost per order and break-even volume
Adds up fixed and variable costs of receiving, storage, picking, packing and returns for your own warehouse and for a logistics operator (3PL), shows the cost of an order and of a unit, the volume at which each option wins and how the difference changes when volume grows or falls.
What is cheaper at our volume: an own warehouse or a logistics operator.
Warehouse flows: hours, staff and load by sale stage (BS, IS, AS, SO)
Calculates labor on the warehouse flows (receiving and putaway, picking and packing, returns) and shows how many people are needed in a normal month and at each stage of a sale period: before the sale (BS), in the sale (IS), after it (AS) and in the final clearance of leftovers (SO). Shows the peak, the staffing gap and the cost of an order at the peak.
How many people are needed for the sale peak and the returns wave after it.
Last mile by delivery type: courier, pickup point, post
Compares delivery types to the customer by the full cost to you: the carrier’s rate minus the customer’s fee, the return leg and the margin lost on refused parcels. Shows the weighted cost of an order, its share of the basket and margin, the lead time and the gain from shifting a share of orders to a cheaper type.
Which delivery type is cheaper for us allowing for refusals.
Split shipments: the cost of several parcels versus consolidation
Shows what it costs to ship an order in several parcels from different places (the second parcel, customer contacts) and compares it with gathering the order in one place (moving goods and the risk of losing the order while waiting).
What is cheaper: ship an order in several parcels or gather it in one place.
Delivery promise and conversion: what pays for a fast promise
Compares options of a promised delivery time (tomorrow, 2–3 days, a week) by contribution to profit: orders at your conversion minus the cost of keeping the promise. Shows the best option and the conversion at which a fast option matches the slow one.
Whether a fast delivery promise pays back.
Warehouse capacity for the peak: places, area, utilization
From the average stock and the seasonal peak coefficient it counts the peak stock, the required number of pallet or bin places at the target utilization and the area with aisles, then with a growth margin. It shows the warehouse utilization in the low season and warns if the utilization at the peak is above 90% when the warehouse work slows down.
How many places and how much area are needed for the peak, not the average stock.
RFID payback: stock accuracy, availability, stock counts
Adds up the annual RFID effect from three sources: the margin from higher goods availability, the saved hours of stock counts and the reduced losses. Compares with the first-year costs (tags, equipment, operation), counts the payback period, the annual effect net of recurring costs and the number of tags at which the investment breaks even. Tags are bought every year for new goods, so the recurring part is shown separately.
Whether RFID tags pay back: the effect, the costs and the time.
Returns processing: cost and restocking
Counts what one return costs: receiving, inspection and packing plus the part of the price that will not come back because some items go to B-grade at a discount and some to scrap. From this it gets the loss per return and per month, the cost of the capital frozen for the time until the return to sale, and how much shortening this time saves.
What one return costs and what shortening the time to restock gives.
Warehouse slotting by ABC: a shorter picking path
Compares the average picking path with a random layout (goods spread evenly over zones) and with an ABC layout (fast goods closer to the exit). From the share of time spent walking it gets the picking time saving, the hours and rubles per year and the payback of the re-slotting costs.
How much the picking path shrinks with an ABC layout.
Self-check questions
- How many orders a month do I usually have and at the peak, and how are they spread by stage?
- What is my break-even volume between my own warehouse and an operator?
- What is included in the operator’s price and what is paid separately: storage, insurance, returns?
- Will people suffice at the peak and after it, when returns come in?
- Which delivery type is cheaper per redeemed order?
- What share of orders ships in several parcels and what does it cost?
- Is the conversion lift from a fast time larger than its cost?
- What claim terms and fines does the operator’s contract contain?
Knowledge check
Multiple-choice questions. The correct answer and the explanation follow each question.
Why can an own warehouse and a 3PL not be chosen by the cost per order at today’s volume alone?
- A. The cost per order at a 3PL is always higher, so an own warehouse always wins
- B. The cost per order does not depend on volume, so comparing contracts is enough
- C. The cost per order cannot be calculated for a 3PL: the operator does not disclose rates
- D. An own warehouse has a large fixed part of costs and a 3PL a variable one, so the better option changes with volume and the crossing point is needed
Correct answer: D. An own warehouse has a large fixed part of costs and a 3PL a variable one, so the better option changes with volume and the crossing point is needed. An own warehouse pays rent, staff and a system regardless of orders, so at low volume an order is expensive and at high volume it gets cheaper. At a 3PL almost everything is variable: the fixed fee is small but every order and storage unit costs money. The cost lines cross at the break-even point: to the left the 3PL wins (and needs no investment), to the right the own warehouse. Lead times, penalties for errors and handling of marking are not in the cost per order: check them in the contract.
Own warehouse: fixed costs of RUB 600,000 a month and RUB 90 variable per order. 3PL: a fixed fee of RUB 40,000 a month and RUB 150 per order. From how many orders a month does the own warehouse become cheaper?
- A. 9,334 orders
- B. 4,667 orders
- C. 10,000 orders
- D. 10,667 orders
Correct answer: A. 9,334 orders. Costs are equal when 600,000 + 90 × N = 40,000 + 150 × N, so N = (600,000 − 40,000) / (150 − 90) = 9,333.3: from 9,334 orders the own warehouse is cheaper. Below that the 3PL wins: at 6,000 orders the own warehouse costs RUB 1,140,000 and the 3PL RUB 940,000. 4,667 comes from adding the variable rates, 10,667 from adding the fixed costs, 10,000 from not subtracting the 3PL fixed fee.
The average basket is RUB 4,000, margin 50%. Post: rate to you RUB 220, the customer pays RUB 150, refusals 12%, return leg RUB 200. Courier: rate RUB 350, the customer pays RUB 150, refusals 3%, return leg RUB 300. Which type is cheaper by full delivery cost and by how much?
- A. The post is cheaper by RUB 130, the difference of rates after the customer’s fee
- B. The courier is cheaper by RUB 65: the post’s refusals eat its rate advantage
- C. The post is cheaper by RUB 65
- D. Both types cost the same: the rate and refusals offset each other
Correct answer: B. The courier is cheaper by RUB 65: the post’s refusals eat its rate advantage. Full cost = rate − customer fee + refusal share × (return leg + order margin); the order margin is 4,000 × 0.5 = RUB 2,000. Post: 70 + 0.12 × 2,200 = 70 + 264 = RUB 334. Courier: 200 + 0.03 × 2,300 = 200 + 69 = RUB 269. The courier is cheaper by RUB 65, although by the rate after the customer’s fee the post is cheaper by RUB 130 (70 against 200). The types are not equal: 334 and 269.
There are 5,000 orders a month, and 12% ship in several parcels. An additional parcel costs RUB 240, the customer contacts support on 30% of split orders and a contact costs RUB 100. Consolidation: moving costs RUB 100 per order and 4% of orders with a margin of RUB 2,500 are lost. How much a month does consolidation save on split orders?
- A. RUB 24,000
- B. RUB 102,000
- C. RUB 42,000
- D. RUB 350,000
Correct answer: C. RUB 42,000. Split orders are 5,000 × 12% = 600. Splitting costs 240 + 0.3 × 100 = RUB 270 per order, consolidation 100 + 0.04 × 2,500 = RUB 200. The saving is 70 × 600 = RUB 42,000 a month. RUB 24,000 comes from forgetting the contacts (240 − 200 = 40), RUB 102,000 from forgetting the lost orders (270 − 100 = 170), RUB 350,000 from multiplying RUB 70 by all 5,000 orders.
There are 100,000 visitors a month who see the delivery time, the basket is RUB 4,000, margin 40% (RUB 1,600 per order). The “4–7 days” promise: conversion 1.8%, keeping the promise costs RUB 100 per order. The “tomorrow” promise costs RUB 350 per order. What conversion does “tomorrow” need to match in contribution?
- A. 1.80%
- B. 2.08%
- C. 2.30%
- D. 2.16%
Correct answer: D. 2.16%. Contribution = orders × (order margin − cost of keeping the promise). For “4–7 days” orders are 100,000 × 1.8% = 1,800 and the margin after keeping the promise is 1,600 − 100 = RUB 1,500; for “tomorrow” 1,600 − 350 = RUB 1,250. They are equal at a conversion of 1.8% × 1,500 / 1,250 = 2.16%: the fast option must win 20% more orders. 2.30% comes from not subtracting the cost of keeping the slow option (1.8 × 1,600 / 1,250), 2.08% from adding the surcharge to the conversion as a share of the margin. Take conversion from your own test.
Why is peak utilization kept not above 90%?
- A. The law requires it
- B. Above it goods cannot fit
- C. Above it processes interfere and the warehouse slows down
- D. It does not matter
Correct answer: C. Above it processes interfere and the warehouse slows down. It is your rule, not a norm. Covered in: A warehouse is counted for the peak, a return by its outcomes.
What is part of the loss per return?
- A. Price minus the restock value plus processing
- B. Only receiving
- C. Only scrap
- D. Only cost
Correct answer: A. Price minus the restock value plus processing. B-grade, scrap and processing are accounted for. Covered in: A warehouse is counted for the peak, a return by its outcomes.
Common mistakes
- Comparing tariffs without storage and returns. Consequence: the “cheap” operator proves dearer than your own warehouse. How to avoid: calculate the full cost per order for a year.
- Counting the peak by the average month. Consequence: no people at the sale, deadlines break. How to avoid: count staff by the busiest week and add hours for returns.
- Choosing delivery by tariff rather than the full cost with non-redemption. Consequence: overpayment and lost margin. How to avoid: divide the expected cost by the redemption share.
- Not building paid storage and receiving fines into a 3PL. Consequence: the month-end invoice exceeds the estimate. How to avoid: request a full calculation with examples.
- Not reading the operator’s claim terms. Consequence: lost goods are not compensated. How to avoid: fix terms, the act procedure and the liability limit in the contract.
Designing a warehouse by the average stock
What it looks like (illustrative figures): The peak breaks the work.
Russia and the EAEU
What matters in Russia. Marking: light-industry goods (clothing, footwear, some accessories) move through a warehouse with Data Matrix codes (a two-dimensional code on the label) of the Chestny ZNAK system. Receiving, storage, shipping and return come with code checks and entries into circulation; Resolution No. 1956 of 31.12.2019 sets the rules and waves are introduced by amendments. Fulfillment operators fix in the contract who scans codes and who answers for errors.
Documents. A handover of goods and warehouse services are drawn up as transfer documents (UPD) or acts in e-document exchange; for storage with an operator a custody delivery note is often drawn up. Taxes: VAT 22% from 01.01.2026 on warehouse and delivery services; if you are on the simplified regime, the operator’s VAT is a cost. Platforms: for FBS and FBO schemes (goods at the platform’s warehouse or the seller’s) receiving rules, terms and tariffs are set by the platform; the platform economy law No. 289-FZ applies from 01.10.2026. We do not make up rates: take them from your contract. Warehouse rents and last-mile rates depend on region and season, check the current ones.
Inside the EAEU there are no customs borders between countries, so a warehouse in Belarus, Kazakhstan, Armenia or Kyrgyzstan can serve several markets, but taxes follow Annex 18 to the EAEU Treaty: VAT at the destination country, an import declaration and supporting documents are needed. As a guide for VAT rates by country are on the page “EAEU and CIS countries” (as of 14.11.2026; check with an accountant).
Marking is introduced on different dates by country: in Kazakhstan mandatory marking starts on 01.12.2026 in stages (a guide, check the text of the order); for Belarus, Armenia and Kyrgyzstan the dates: indicative as of 14.11.2026, check the text of the act. For goods moving between countries check the codes and the operators' roles. Take logistics operators' rates by country from the operators' offers.
Categories and audiences
- Apparel
- On apparel: many SKUs by size and season, picking takes longer, returns after fitting are high. Hangers and garment covers need space. Count hours by pieces per order, not by number of orders.
- Footwear
- In footwear: boxes take a lot of space, weight and volume are higher, so storage and delivery cost more per unit. Returns by size are frequent and checking a pair and repacking is labor-intensive. Allow more space and hours for returns.
- Accessories
- For accessories: small and valuable, high storage density, but a higher risk of loss and mixed stock. They need unit-level tracking and receiving control. Delivery by post or to a pickup point is often cheaper than courier because of low weight.
- Kids
- If it is children’s wear: a peak before 1 September and the New Year, many small sizes and ages, a high share of urgent orders. Check marking and conformity documents on receiving, a mistake here costs more. Allow a staff reserve for the last two weeks of August.
- Men
- In a men’s warehouse load is set by bulk: outerwear and footwear take the volume of several dozen T-shirts. So warehouse capacity is counted in cubic meters and pallet places, not in units, and the shipping peak of outerwear is planned separately by groups. A single count in “units per person” misleads here.
- Women
- The bottleneck of a women’s warehouse is the return: an item after fitting comes back with the tag but without packaging, and it must be inspected, steamed and returned to sale. Count the hours per returned item and the share of returns that goes back to stock rather than to markdown: the staffing after sales depends on this.
Questions this topic answers
- Own warehouse or an operator, and from what volume does the answer change?
- Are there enough people for the sale peak and the returns after it?
- Which delivery type is cheaper allowing for refusals?
- Does a fast delivery promise pay for its cost?
Key measures and terms
Study first
What next
Related: “Replenishment and stock cover”, “Stock accuracy, shrink and availability”, “Channels: wholesale, retail, platforms”.
Documents on this topic
- Warehouse: WMS, handling cost, SLA, returns and cycle counting
- Omnichannel: OMS, ship-from-store, BOPIS and unified stock
- Loss prevention: shrinkage, counting, fraud
- Accounting for defects, mis-sorts and stock losses: acts, liable parties, corporate income tax
- Warehouse and fulfillment: slotting, order picking, capacity and peak, KPIs, own warehouse or a logistics operator
- Business insurance: cargo, property, liability, receivables, how to read a policy and when insurance pays off
- Reader question: a container or consolidated cargo for a batch of 800 garments and what demurrage is
- Warehouse and data: capacity for the peak, stock and sales reconciliation, returns, RFID and slotting