Food and Beverage Distribution Business: Startup and Operations Guide

Food and beverage distribution is a replenishment business. Independent c-stores, grocers, and restaurants reorder weekly if you do not stock them out or serve them warm yogurt. Recurring drops fund the van. The tax is dating, temperature, and recall speed. US operators should assume they need a food facility registration mindset, a state wholesale food license where the state requires one, sanitation SOPs, lot tracking, and a written recall plan before a serious retailer will cut a vendor number. Alcohol is not “grocery with a higher margin”: it sits in the TTB three-tier system plus state wholesaler permits.

Directional margins: specialty beverage and specialty grocery often land 18–28% invoice gross after inbound freight if spoilage stays under 2% and you are not buying closeouts you cannot date. Broadline-style commodity cases can sit closer to 10–16% and only work on dense routes. If your plan needs 35% gross on milk or national soda, you do not have a plan — you have a hope. Price the route, not the case.

Register the facility with FDA when you hold food for sale in the US — see FDA food facility registration. Then map every SKU against the Food Traceability List in the FSMA 204 final rule. If the SKU is on the list (or you handle it in a way that triggers the rule), you need Key Data Elements and Critical Tracking Events — not a spreadsheet named “lots-final-FINAL.” The operational test is ugly and simple: a buyer or regulator names a lot at 10:14 a.m. and you say where it came from and which invoices it shipped on before lunch.

State wholesale food licenses, local health inspections, and allergen/sanitation SOPs still apply even when FDA registration is done. Cold-chain SKUs need storage and delivery temperature controls with logs you can produce — continuous probes or a validated spot-check program, not a driver who “usually remembers.” Use FEFO (first expired, first out) for dated product. FIFO is a warehouse habit; FEFO is a dating habit. A pallet of yogurt put away by receipt date in front of an older lot is how you donate margin to the dumpster.

Territory: one metro, 40 independent c-stores already buying premium beverages from a sleepy competitor. Line card: 18 SKUs (4 brands, 2 pack sizes). Opening inventory: $48,000 at cost — about 3.5 weeks of forecast, not 12. Van: one refrigerated sprinter, $1,100/week fully loaded (payment, insurance, fuel, driver). Terms: you pay suppliers Net 15; stores pay you weekly COD or Net 7. That gap is the cash you must fund.

Week-8 run-rate if execution is clean: 40 doors × 1.8 drops/week × $165 average drop = $11,880 weekly invoices. Invoice gross 23% = $2,732. Spoilage and dating 1.6% of sales = $190. Route cost $1,100. Contribution about $1,440/week before occupancy and owner draw. Miss dating (3.5% spoilage) or let average drop fall to $95 and contribution vanishes. That is why the first hire is not a “brand ambassador.” It is someone who counts lots and holds a cutoff.

Write the recall SOP before the first PO. Name the coordinator, the after-hours phone, and the system of record for lot → invoice → door. FSMA 204 is pushing the industry toward hours, not “we’ll look next week.” Run a mock recall quarterly: pick a real lot, time the traceback and the list of doors, and fix the scan gaps you find. If your WMS cannot print that list, you are not ready for a chain that will fine you for being late.

Receiving is where recalls are won. Refuse product without a readable lot and a temperature on refrigerated inbound. Photograph the seal and the logger. Put away to FEFO locations, not “the empty bay by the door.” When a supplier later says lot 1842 never left their plant, your inbound photos are the argument. CDC outbreak pages are not your SOP, but CDC foodborne outbreak resources are a reminder that your customers will hear the news before your supplier’s legal team finishes a statement.

Start with a buyer problem, not a brand crush: “premium non-alc for independent c-stores inside I-285” beats “we carry beverages.” Negotiate MOQs you can turn in 21–28 days. A pallet MOQ on a 90-day code date is how you fund the landfill. Ask for dating guarantees, spoilage allowances, and price protection if they cut a promo two weeks after you buy. Off-invoice “support” that requires you to hang tags in 40 stores is labor — price it.

Kill SKUs weekly. If a flavor has not moved in 21 days and the code date is inside 45, mark it down on the next route or return it under the allowance. Vanity variety is the enemy of FEFO. Your A-list should be the SKUs that appear on 70% of invoices. Everything else is a science project until a buyer proves a second drop.

If you add beer, wine, or spirits, stop treating the SKU like grocery. US alcohol wholesale is a state-licensed, often three-tier business. TTB and the state ABC set who may import, warehouse, and sell. Crossing a state line with a sprinter full of IPA is not a “route extension.” It is a permit question. Build alcohol as a separate company or a separately licensed division if your grocery license cannot carry it.

Imported specialty food still needs a US facility/importer identity, prior notice where applicable, and a lot story that survives FSMA 204. A pretty tin from a trade show with no English label plan and no responsible US party is not a catalog item. Landed cost must include duty, freight, aging on the water, and the labor to relabel. If that math only works at a 40% margin you will never get from a c-store, pass.

The 40-door sparkling-water book dies in the slot, not on the route sheet. Refrigerated SKUs get a dedicated face ordered by expiry, not by receipt date. Ambient cases get a dry aisle that does not share a pallet jack path with the cooler so doors stay shut. Quarantine sits at receiving: no lot, no temperature, no put-away. Yellow-tag anything inside 21 days on a 60–90 day juice code so a picker cannot bury it behind a pretty new pallet. Overflow “deals” from a brand do not land on the A-face. They land in a promo bay with a kill date. If you only have one bay, you do not have a deal — you have a dating problem with a discount on it.

Load the sprinter the same way you slot: FEFO, cold first, invoices that match lots. A tote of mixed dates “to make the drop faster” is how a mock recall takes 6.5 hours instead of 2. Stage by stop order after you stage by date. Returns from stores — code-dated, warm, or damaged — go to the cage, not back to the face. Photograph and log. The $48,000 opening inventory only turns if the oldest sellable unit is the one the picker sees.

You paid Net 15 on that opening book. If 40 stores pay weekly COD, cash conversion is almost honest. If half of them talk you into Net 14 “like the last guy,” you are funding another two weeks on $11,880 of weekly invoices — call it $24,000 extra float — plus the $190 of spoilage you already modeled. A chain that wants Net 30 and a 2% shortage deduction is a different product than an independent that pays the driver. Do not mix them on one credit policy. COD or card until three clean weeks. Then a small limit. Personal guarantee on thin independents who want a cooler full of consignment they will not call consignment.

Age AR twice a week, not at month-end. Food buyers forget a drop that is more than ten days old. The messy week in the table ($7,600 invoices, $360 contribution) gets worse if $2,200 of that is 40 days past due and you still reload them. Stop the next drop at 21 days. Sales will call it harsh. Finance will call it still being in business. If a supplier will not move off Net 15 until you have history, fund the gap with owner cash or a small line — not by stretching dating until spoilage hides the cash hole.

Do not open with a brand story. Open with: I can be here Tuesdays and Fridays before 10, cutoff is 3 p.m. the day before, these 12 SKUs are in date, and if a lot is bad I can tell you who else got it before lunch. Show one receiving photo and one invoice with a lot. Ask what they stock out of every week and what they throw away. Write both. Your first order should refill the stockout, not the launch display the brand sent you. If they want 18 flavors and you have a 90-day code, sell six and a restock promise. The sleepy competitor loses on missed drops and warm coolers, not on your brochure.

When they ask for a cooler on loan, price the labor and the power and the exclusive-facing they will not give you. When they ask to return anything that does not sell, point at the 21-day / 45-day kill rule and the spoilage allowance you already negotiated — not an open dump back on the van. When they show you an Amazon six-pack cheaper than your case, do not match. Check whether it is the same GTIN and whether that seller is authorized. Specialty beverage MAP is weaker than electronics, but a dumped pallet still trains the next 39 doors. Document and call the brand.

It fails when average drop slides from $165 to $95 because you added 30 doors to look busy. It fails at 3.5% spoilage because FIFO hid the old lot. It fails when a mock recall cannot list doors in four hours and a chain fines you or cuts the vendor number. It fails when alcohol cases get on the grocery van without a TTB/state story. It fails when you buy a pallet MOQ on a 90-day code to hit a rebate. The clean week in the table ($1,440 contribution) is not a personality trait. It is cutoff, FEFO, and 40 doors you can actually stand in twice a week.

A slower fail is the science-project SKU that never hits 70% of invoices and still occupies cooler face. Kill it on a schedule. Another slow fail is a driver who “usually remembers” the probe. One warm load is a complaint file and a dumpster. Run the Wednesday recall drill even when sales is busy. If the WMS cannot print lot → invoice → door, you are not ready for the account that will ask. That is a control problem, not a branding problem.

Do not buy a food WMS because a demo showed a pretty lot screen. Run the mock recall on paper or in the ERP you have. If you can name the lot, the inbound photo, and the invoices before lunch, you may only need barcode put-away and FEFO location flags. If the drill takes a day and three CSVs, then buy lot/expiry and receiving capture — not a full commerce suite. EDI comes when a named grocer makes 850/856/810 a gate. Temperature logging can start as validated spot checks with a binder. Continuous probes are for when a buyer or an insurer asks, or when you add ice cream.

Skip route-optimization theater until 40 doors are dense and contribution is boring. A $1,100 sprinter week is lost in software subscriptions faster than founders admit. When you do implement, implement receiving and FEFO first, invoicing lots second, catch-weight or variable case only if the SKU requires it. If a vendor leads with “AI replenishment” and cannot show a lot-to-door print, keep walking. FSMA 204 is a record problem. Buy the record.