Choose a fulfilment partner using a peak-day operating test

Two fulfilment partners quote similar prices per order. One demonstration processes simple single-item orders quickly; your promotion includes bundles, fragile products and a sharp carrier cutoff. The quoted unit price does not describe the operation you need.

Run a peak-day test that follows the actual order mix from accepted demand to correct carrier handover. The objective is to identify which partner can complete the commitments, explain exceptions and produce a reproducible bill.

Give every candidate the same test population

Prepare a controlled set of orders with item dimensions, units, packaging rules, bundles, delivery services and cutoffs. Include opening inventory, holds, replenishment needs, returns and work already in progress.

Freeze the population and its original promises. If a partner excludes an order as unsupported, retain that outcome in the comparison. Otherwise the apparently faster partner may simply process an easier subset.

Define the expected evidence: receipt and inventory records, pick and pack confirmations, handling-unit identities, carrier handover, exception decisions and billing lines. A demonstration should produce usable records, not only a summary slide.

A peak with sixty-five hours of stated work

This illustrative peak contains 1,200 orders: eight hundred simple orders, three hundred bundles and one hundred fragile orders. The agreed teaching estimates cover the specified pick-and-pack tasks only.

The workload is 3,900 labour-minutes, or sixty-five hours. Eight people with 360 usable minutes each provide 2,880 minutes, or forty-eight hours. The shortfall is 1,020 minutes, or seventeen hours.

Eleven equally eligible people would provide 3,960 minutes, leaving only sixty minutes against this task estimate. That arithmetic does not prove feasibility: receiving, replenishment, supervision and unplanned exceptions are excluded, and station or equipment limits may constrain parallel work.

Ask each partner to show the time-phased roster and the resources behind it. Three extra people do not help fragile packing if only one qualified station can perform that work.

Make the operating test include disruption

Use one bundle with a missing component, one held lot, a carrier cutoff change, a label-service interruption and a customer cancellation after picking. The partner should show how each case is controlled and how the remaining work is replanned.

Measure the complete path. A high pick rate can coexist with a packing queue that misses handover. An “order shipped” status should have a defined physical carrier event, not merely a printed label.

Observe the work or review evidence from a controlled trial. A proposed staffing plan and a live result are different levels of evidence; label them accordingly.

Reproduce the commercial model

Ask for the charge grain and minimum for receiving, storage, picking, additional items, packaging, returns and value-added work. Define which activity triggers a charge and how corrections are treated.

CartonCloud advertises storage, handling and other activity-based billing capabilities. That illustrates why the tariff and activity records must be evaluated together; it does not validate any partner's quoted savings. CartonCloud — Billing and invoicing

For an illustrative comparison, Partner A charges ₹25 per order plus ₹8 for each of the three hundred bundle orders and ₹15 for each fragile order. Its stated service total is ₹30,000 + ₹2,400 + ₹1,500 = ₹33,900. Partner B charges a flat ₹29 per order for the same defined scope, totalling ₹34,800.

The ₹900 difference is meaningful only if packaging materials, storage, returns, minimums and excluded services are genuinely comparable. Mark unknown charges as unresolved rather than zero. Do not assume a per-order price covers every item or handling step.

Inspect inventory and data boundaries

Confirm how customer-owned stock, units, lots and returns are separated. Ask the partner to reconcile a sample item from receipt through a split, shipment and return. A correct total across all clients is insufficient.

Test portal and API access with authorised sample accounts. OWASP's object-level authorisation guidance applies to orders, stock, POD files and downloadable reports. OWASP — Broken Object Level Authorization

Require an export that preserves the identifiers and relationships needed to move to another provider or investigate a dispute. A spreadsheet of latest statuses may omit the history that explains inventory and billing.

Compare the evidence consistently

Score essential controls as gates, then compare service, effort and price among candidates that pass them. Record each result as demonstrated, proposed, dependent on configuration or unresolved.

Use the original 1,200-order population for completion and correctness measures. Show late, cancelled, held and unserved orders separately. Compare actual labour and exception effort with the plan, including any external support added for the demonstration.

A partner that admits a capacity gap and offers an explicit staged plan may be more predictable than one that reports complete fulfilment without handover evidence. The decision should rest on what can be sustained under the agreed scope.

Choose a partner for the difficult operating day

The useful test connects order mix, resources, exceptions and charges. It gives procurement and operations the same basis for selecting a service they can explain after the peak.

Next actions

1. Freeze a representative peak population and its commitments.
2. Require a resource plan covering the full operating flow.
3. Execute disruption cases and reconcile all order outcomes.
4. Recalculate the bill from the observed activities and agreed tariff.

Bring the proposed peak mix and the current exception backlog to a fulfilment-partner assessment; average orders per day will not expose the same constraints.