Logistics
Set inventory policies that protect fill rates while reducing excess stock and cash tied up in slow-moving items.
An item that hits 12 times a week at $2 GP$ is not your A item if a $40 GP$ item hits twice and stocks out. Rank on GP$ at risk: frequency × margin × stockout pain (the doors that will dual-source you). Classic unit-ABC puts candy in A and the $180 gasket in C. The gasket closes shops. The candy does not.
Give A items a service target (97–99% fill), a weekly count, and a buyer who answers the phone. B items get a monthly review. C items get a min-max you will actually cut, and a kill date if they do not move in 120 days. One policy for 2,400 SKUs is how you fund a museum.
Seasonal SKUs are a fourth letter. Halloween is not a C item in August. Tag the calendar. Safety stock that sits from November to September is a loan to last year's forecast.
ROP = (average daily demand × lead time in days) + safety stock. Safety stock ≈ Z × σdemand × √lead time if you want the textbook, or a simpler 'cover the worst week you have seen in 6 months' if your data is dirty. Dirty data plus a fancy formula is a confident stockout.
Worked ROP. Item 4410, 8 cases/week average, σ about 3 cases, domestic lead time 10 days, you want ~95% (Z ≈ 1.65). Daily demand 8/7 ≈ 1.14. ROP ≈ 1.14×10 + 1.65×3×√(10/7) ≈ 11.4 + 5.9 ≈ 17 cases. If the buyer is still using a min of 6 because 'that is what we always did,' you will stock out every time a week runs hot.
Imported items replace 10 days with 45–70 and a worse σ. That is why a 'cheap' FOB is expensive: you buy safety stock you can see on the balance sheet. Recalculate when the supplier's on-time slips. A static min-max through a Red Sea delay is how you explain shorts to a banner.
Turns = COGS / average inventory. Days on hand = 365 / turns. A house at 6 turns is sitting about 61 days. Moving to 8 turns frees a lot of cash and will cost fill if you cut A-item buffers to get there. Cut C items first. Anyone who promises both 12 turns and 99% fill on a long-tail catalog is selling software.
The Census AWTS inventory-to-sales figures tell you whether your days on hand are eccentric for the kind of business. Use them in the bank conversation. Do not use them to fire a buyer who is protecting a 98% A-item fill in a messy lead-time year.
Publish fill and turns on the same page. A buyer who is a hero on turns and a villain on A-item fill is not a hero. A buyer who is 90 days fat and 99% fill is a working-capital problem. You want the pair.
At 90 days with no movement, flag. At 120, assign a fate: return to supplier, liquidate, bundle, or write down. At 180, it is a religion. Religions do not pay rent. Give sales a 30-day exclusive to move it at a published price, then it goes to the listing or the recycler.
Worked leftover. A 'deal' buy: 1,200 cases at $18, save $2.10 versus everyday. You sold 520 in the first 90 days and 80 in the next 90. Remaining 600 cases × $18 = $10,800, plus 180 days of capital and a half-aisle. The $2.10 save on 520 cases was $1,092. You lost. Deal buys need a pre-sell or a kill.
Returns to supplier only work if the terms say so. Negotiate a defect and a slow-mover window on the way in. Begging on the way out is not a policy.
A 12-tab workbook is where SKUs go to hide. Give the buyer a daily list: A items below ROP, incoming POs that will miss, and C items over 150 days. That is a morning. The rest of the catalog can wait for the weekly pass.
Substitutions need a list too. If 4410 is short, what ships, at what price, and does the customer allow it? Unapproved substitutions are chargebacks. Empty substitutions are shorts. Write the pair before peak.
Count what the system thinks you have. A weekly A-item cycle count and a monthly B pass will do more for fill than a new planning module on top of a 6% location error. Inventory policy on a lie is just a smoother lie.
CCC = DIO + DSO − DPO. The buyer owns DIO. If they stretch DIO to protect fill while sales stretches DSO to win doors, you will visit the bank. Put CCC on the same whiteboard as fill. The SBA note on managing finances is generic and still right: growth that outruns cash is a failure mode, not a phase.
MOQs fight DIO. A 50-case minimum on an item that sells 6 a week is 8 weeks of stock before safety. Either negotiate a mix, pre-sell, or refuse the line. Do not hide the MOQ inside a 'strategic brand' story.
Transfers between buildings are not free replenishment. They are freight plus two receives plus a chance to lose the lot number. Transfer A items to protect a promise. Do not transfer C items because a satellite manager is nervous.
A supplier offered a 90-day dating and a 6% off bill-back if you took the full color run. The buyer shook. The color run was 14 SKUs, 8 of which your doors had never ordered. Dating expired. Bill-back required a scan-back you did not have. You owned $64,000 of pretty boxes and a fill problem on the two SKUs that actually moved, because the aisle was full.
Rule after that: no deal buy without a pre-sell list or a return window in writing, and no bill-back in the sell price until the claim files. Handshakes are how warehouses get interesting.
Show the aging in dollars, not SKU counts. Owners react to $64,000. They do not react to 'we have some slow movers.'
If the category has a date — food, some chemicals, medical adjacent — FEFO is not a preference. Slot by date, pick by date, and refuse a 'customer wants the new lot' override without a manager code. That override is how you own expired cases in June that you received in January.
Supplier date games show up as short-code inbound. Write a minimum remaining shelf life on the PO — 75% of life, or 120 days, whatever the category needs — and reject at the dock. A 10% 'deal' on 40-day remaining life is not a deal if your doors turn the item in 50.
Lot traceability is a recall muscle. You should be able to say which doors got lot 24A03 in 30 minutes. If the answer is 'we would have to look at invoices,' you do not have traceability. You have a hope. FDA recall reality does not wait for your hope, even if you are 'just the distributor.'
Vendor-managed or consignment stock still occupies a location and a count. If it is in your building and your system cannot separate ownership, you will sell it twice or you will not sell it when you should. Tag ownership. Cycle-count it. The supplier's 'we'll watch it' is not a count.
In-transit inventory is DIO you cannot pick. A PO that has been 'shipped' for 11 days is not safety stock. Do not let the buyer talk about it as if it were on the floor. Split on-hand and on-order on every A-item view.
Customer-reserved stock — 'hold 40 cases for Banner X' — is how A items disappear for everyone else. Time-box the reserve (72 hours, or until the PO posts) and charge a hold fee if they flake. Unscoped reserves are how independents learn you always short them for the logo.
Written by
James Cole is a wholesale operator who has run distribution P&Ls through first-warehouse launch, inventory turns, trade credit, and EDI-backed accounts.
Published April 29, 2026 · Last reviewed July 15, 2026
Continue reading in this topic or explore a related distribution guide.

Logistics
Learn how to run warehouse, inventory, fulfillment, freight, and transport operations that improve fill rate and reduce total landed cost.
March 17, 2026
Logistics
Design warehouse operations that increase pick productivity, reduce errors, and support multi-channel B2B fulfillment.
April 24, 2026
Logistics
Understand when to use 3PL providers, how to compare options, and how to manage service-level agreements effectively.
July 10, 2026