Order routing decides which location is allowed and best positioned to fulfill demand. A low-cost location is not useful if it cannot ship the product, meet a channel restriction or handle the requested service. The safest design filters for eligibility first and optimizes only within the remaining choices.
Define hard eligibility gates
List conditions that make a location eligible for a specific order or line. Common gates include:
- sellable and reservable stock;
- facility capability for the product, pack or service;
- channel, seller or geography authorization;
- hazardous-material or carrier restrictions;
- delivery-promise feasibility;
- operational availability and cutoff status;
- allocation or account commitments.
Treat these as facts with sources and timestamps. A location’s inventory snapshot may be too old to prove eligibility. A carrier service can be available generally but not for a particular postal code or product.
Optimize after filtering
Within the eligible set, rank choices using transparent objectives such as expected delivery, variable fulfillment and shipping cost, split count, capacity or inventory-balancing policy. Define tie-breakers so the same inputs produce an explainable result.
Avoid combining every concern into one opaque score. Operators should be able to read: “Location B was selected because A lacked eligible stock; B and C could meet the promise; B had lower modeled cost.” This makes exceptions and audits practical.
The order lifecycle model provides the handoff states after a route is chosen. The shared inventory article explains why source quantity and reservations must be defined consistently.
Work through a hypothetical route
An order contains two units of SKU X for delivery by Friday. Location East shows three units, but one is quarantined and one is already reserved. Location Central shows four sellable units and can meet Friday. Location West has lower handling cost but its ground transit misses the promise.
East is ineligible for quantity two, and West is ineligible for the service promise. Central wins even if its modeled cost is higher. If a split could use one unit from East and one from Central, the OMS should compare that option only if both locations can reserve their lines and the split meets policy.
Reserve and acknowledge the decision
Routing without reservation can create a race. Use the system’s supported reserve-or-confirm sequence, and define what happens when stock changes between evaluation and commitment. The fulfillment destination should acknowledge the handoff before the OMS treats the route as operationally accepted.
Store rule version, input snapshot, candidates, rejection reasons, selected location and acknowledgment. This evidence supports service decisions and later tuning.
Control overrides
An override should require a reason, actor, time and expiration or completion condition. Revalidate eligibility before applying it. Do not let an operator force a location that lacks a legal or physical capability merely to clear an exception.
Useful overrides might hold an order, select another eligible location or permit an approved split. Keep the original automated recommendation for audit.
Recover from routing failures
If no location is eligible, expose the exact gates that failed. Route the issue to inventory, address, product-compliance or capacity owners rather than blindly rerunning the same rules.
If the selected facility rejects the handoff, confirm whether it created work before rerouting. Release the original reservation only through the owning system, then evaluate fresh inputs. This prevents simultaneous fulfillment at two sites.
Accept the tradeoffs
Cost-first rules can erode service; speed-first rules can create avoidable expense and fragmented inventory. Highly dynamic routing reacts to current conditions but becomes harder to reproduce. Use explicit priorities and log the inputs used for each decision.
A useful next step
Sample twenty recent routes. For each, reconstruct eligible candidates and the deciding rule. Investigate any choice that cannot be explained from retained evidence.
Review routing performance by reason, not only average cost. Compare orders that had one eligible location, several candidates, a split, a manual override or no eligible route. This separates optimization quality from availability problems and helps the team change the right rule instead of tuning a score that never had a real choice.
1Commerce can support multi-location order and fulfillment workflows depending on integration configuration and scope. Explore the multi-channel platform or ask 1Commerce to review routing controls.