How to choose a 3PL that fits your business

Define the operation before you shop

Share a clear picture of your products, order volume, sales channels, and busiest weeks. Include the unusual work too—bundles, retail labels, returns, or special packaging.

1. 01Order profile by month for the last 12 months: total orders, total units, and the distribution of units per order. Not the average. The distribution.
2. 02Package profile: your five most common outbound carton sizes with real dimensions and real billable weights, plus the share of orders that ship in each.
3. 03Your product range and bestsellers: how many products you stock and which account for most orders.
4. 04Channel mix: DTC by storefront, marketplace volume by marketplace, and wholesale or retail PO volume broken out by retailer, with routing guide requirements named.
5. 05Seasonality shape: your peak week volume as a multiple of your median week, how many weeks the peak lasts, and whether it is one spike or a plateau.
6. 06Value-added requirements: kitting, bundling, inserts, gift notes, serial capture, lot and expiration tracking, ticketing, poly bagging, FBA prep, and how often each occurs.
7. 07Inbound profile: containers or LTL loads per month, floor-loaded or palletized, case pack counts, whether cartons arrive with scannable labels, and your typical PO line count.
8. 08Growth curve: what the next 12 months look like under your plan case, not your pitch deck case, with the channel that drives the growth named.

The distinction that matters most: parcel warehouses and pallet warehouses

Parcel orders and retail pallets need different processes. If you sell through both, ask to see how the warehouse handles each and keeps the inventory records together.

Evaluation criteria that predict success, and the ones that do not

Look for experience with orders like yours. Receiving speed, accuracy, communication, and flexibility matter alongside the quoted rates.

• Receiving turnaround, stated as dock-to-available in hours or days, measured from trailer arrival rather than from whenever they choose to begin. If receiving takes three weeks, inventory you already paid for cannot be sold, and no software feature compensates for that.
• Inventory checks: how often products are counted, who investigates differences, and who approves changes to the records.
• Handling problems: if an item is missing during packing, who decides whether to split or hold the order, and how are you told?
• A named day-to-day contact, their account load, and the escalation path after 6pm in November.
• Whether they will say their own order accuracy number out loud, and whether they can tell you if it is calculated per order or per line. Per-line accuracy always looks better. Both are fine. Vagueness is not.
• The peak labor model: how early they hire, what share of peak headcount is temporary, and how many of last peak’s pickers came back.
• How value-added work is priced and measured. Per unit, per touch, or hourly against a standard time study, with the standard written down.
• Financial and operational stability boring enough to be uninteresting: how long they have run the building, whether the leadership you are meeting will still be there.

• The customer portal demo. Dashboards are cheap to build and tell you what already happened. The portal will not receive your containers.
• The logo wall. Large brands often occupy dedicated space with dedicated staff, which tells you nothing about how your account will be run.
• Total square footage. Available, slotted, staffed capacity in the specific building your goods will live in is the number that matters.
• A long integration list. Depth of one integration beats breadth of thirty. Ask how inventory syncs, how often, what happens on a failed sync, and who notices.
• The address in isolation. Location genuinely affects transit zones and parcel cost, but a good zone does not rescue a building that cannot receive you in a week.
• The sales engineer. Charming sales engineers are distributed randomly across good and bad operations.

Nile’s order accuracy is 99.98%. When comparing providers, ask how each measures accuracy and which reporting period it covers.

A weighted scorecard you can actually use

Agree which factors matter most to your team before reviewing proposals. Use the same scorecard for every provider and record the evidence behind each score.

The questions that get real answers

Ask for examples of how the team handles receiving delays, inventory differences, and promotional peaks. Practical answers tell you more than a general promise.

1. 01How long did receiving take last month, and during your busiest recent week?
2. 02What is your order accuracy, how do you calculate it, and what happens financially when you ship the wrong thing?
3. 03How often do you count inventory, who investigates differences, and what are your latest accuracy results?
4. 04What is your utilization in this building right now, and what does the third week of November look like here?
5. 05Who would be my day-to-day contact, how many other accounts do they carry, and who covers when they are out?
6. 06How do you staff peak? When do you start hiring, what share is temporary labor, and what was your return rate from last peak?
7. 07What is your warehouse software, who administers it, and what can I configure versus what requires a ticket to you?
8. 08Which customers left you in the last 18 months, and what was the reason? An honest answer here is worth more than three glossy case studies.

• How do you find out an order is a rush?
• What happens when an item is missing while packing an order? Walk me through a recent example.
• How often does anyone count this aisle, and who tells you to do it?
• Where is the returns backlog sitting right now?
• What is the last process change that made your day easier, and who decided it?
• What is the most common reason this building misses a cutoff?

The site visit: what to look at on the floor

A warehouse visit helps you connect the proposal to the work. Follow an order from receipt to shipment, and ask the team to explain the checks along the way.

• Housekeeping in the aisles, not the lobby. Are pallets squared in their locations, is trash contained, are floor markings still visible and respected?
• Dock congestion. Count trailers waiting and look at whether receiving lanes are stacked with unprocessed freight. A receiving lane stacked three deep is a preview of your own inbound sitting unsellable.
• Returns staging. Is there a defined area with a stated process and a person assigned, or a corner where boxes accumulate? Returns are the first thing that slips when a building is behind.
• Visible cycle counting. Is someone counting right now, with a scanner, against a posted schedule? A cycle count program that only exists in the SOP binder is not a program.
• Rack and label condition. Torn location labels and bent uprights are a maintenance posture, and maintenance posture correlates with inventory accuracy more than anyone admits.
• Picker pace. Steady or rushed? Look for people waiting on equipment, carts staged without operators, or pick paths that double back.
• Label and pack stations. How many, how many are down, and is anyone queuing at one?
• The exact zone your goods would occupy. Ask them to walk you there. Hesitation at this request tells you capacity is tighter than the proposal suggested.

References: ask for your profile, then ask about the worst month

Ask to speak with a customer whose products and order mix resemble yours. Discuss everyday service as well as how the provider handled a difficult week.

1. 01Tell me about the worst month you have had with them. What broke, how did you find out, and how long did it take to fix?
2. 02When something goes wrong, do they find it or do you find it from a customer email?
3. 03What happens to the invoice when your volume spikes? Were there line items you did not expect?
4. 04How many times has your day-to-day contact changed, and what happened to service each time?
5. 05What do you wish you had negotiated before signing?

Comparing proposals that are not comparable

Give every provider the same order and inventory data. Compare a full month of costs, including storage, receiving, packing materials, freight, and additional work.

1. 01Export 30 representative days of orders with units, SKUs, shipping destinations, and carton assignment. Export your peak week separately.
2. 02Ask every candidate for a complete rate card with each unit of measure defined in writing, including accessorials, minimums, and any fuel or peak surcharge mechanism.
3. 03Apply each rate card, line item by line item, to your own order file. Do this in a spreadsheet you control, not in their model.
4. 04Produce three outputs per candidate: cost per order, cost per unit, and total monthly cost. Then rerun at double your current volume and at your peak week shape.
5. 05Model storage separately, using your actual delivery schedule. Storage cost depends on how long inventory sits, which depends on their receiving speed and your buying pattern, not on the rate alone.

Questions to resolve before choosing

Follow up when a proposal leaves responsibilities, fees, or service measures unclear. A useful partner should be able to explain how the work will run.

• No written service standards. If receiving turnaround, order cutoff, ship-by standard, and accuracy are not in the agreement with numbers attached, you have no basis to escalate later beyond goodwill.
• Vague units of measure in pricing. “Pick and pack: from $X” or “storage: competitive” means the invoice will be defined after you are dependent.
• No named implementation owner. If the salesperson is still the main contact after signature, onboarding will be run by whoever is free.
• They cannot state their own accuracy number. Every operation that measures accuracy knows the number. Not knowing it means it is not measured, which means it is not managed.
• Reluctance about a site visit, or steering you to a building your account would not run in.
• Pricing that only works at volumes you do not have yet, or a contract term long enough to outlast your patience with a notice period that protects only them.
• Attractive storage rates paired with vague receiving and accessorial pricing. That is a pricing structure designed to win a spreadsheet comparison, not to describe your bill.
• Every question routed through the salesperson. If you cannot get operations leadership on a call during the sales process, you will not get them during a crisis.
• They cannot name their warehouse software, or it is homegrown and undocumented. You should know the platform. Nile runs Deposco, for instance. The answer matters less than the existence of an answer.
• No stated process for what happens when they are wrong. Mis-ships happen everywhere. A provider with no claims process has simply not decided who pays.

Implementation: what a real onboarding plan contains

Agree on data setup, connection testing, inventory transfer, and launch responsibilities. Your promotions and current agreement should help shape the timeline.

1. 01A named implementation owner with a phone number, plus how many other go-lives that person is running in the same window.
2. 02A dated milestone plan built backward from a go-live date both sides signed, with dependencies marked.
3. 03Product records: every SKU with dimensions, weight, case pack, barcode, lot and expiration flags, hazmat classification where relevant. Whoever owns this owns your future accuracy.
4. 04Integration build and test using real order data, including the ugly cases: multi-line orders, backorders, address failures, gift notes, subscription rebills, marketplace orders, and returns.
5. 05An inbound plan with dated dock appointments, not “send us your inventory.”
6. 06A cutover approach with a rollback: channel by channel or hard switch, and what happens if day three goes badly.
7. 07A first-week escalation path with names, cell numbers, and hours.
8. 08Written success criteria and a dated 30-day review against them.

The decision framework

Choose the team and operating model that fit your business. Record the assumptions, confirm the scope, and make sure both teams know the next steps.

1. 01Build the profile document. Do not contact anyone until it exists.
2. 02Longlist six to eight providers and screen to four on profile fit alone, using the parcel-versus-pallet question and the volume split.
3. 03Send the same data file to all four and ask for a complete rate card, not a bundled package price.
4. 04Model every proposal against your own order file at current volume, double volume, and peak week shape.
5. 05Visit the specific building your goods would occupy, during a busy window, with your ops lead present.
6. 06Take two references each, matched to your profile, and ask the worst-month question.
7. 07Score the weighted scorecard independently, then reconcile disagreements out loud.
8. 08Negotiate the exit before the entry: notice period, transfer-out pricing per pallet, data export format and timing, ownership of the product records, and a defined right to terminate for repeatedly missed service standards.
9. 09Pilot where you can. One channel, one SKU family, or one region first, with a defined go or no-go date.
10. 10Book the 90-day review at signature and review against the exact numbers you scored.