Research library

Fulfillment operations

B2B Inventory Availability: Reservations, ATP, and Reconciliation

A practical operating guide to sellable stock, order reservations, available-to-promise rules, and inventory reconciliation across commerce systems.

Sources checked: October 17, 2026

Define the availability promise

Start with the quantity a buyer or operator is trying to understand: what can be promised for a specific item, location, channel, and time? Record the SKU or variant, unit of measure, site or fulfillment location, applicable account or channel scope, inventory state, and the time the value was calculated. An on-hand total, a sellable quantity, and a date-specific available-to-promise (ATP) result can answer different questions. Define those terms for your own workflow before connecting systems or exposing a quantity to buyers.

Separate physical, sellable, reserved, and incoming stock

Write down how the operating model treats physical on-hand stock, quantities committed to open orders, unavailable or held stock, inbound receipts, safety buffers, and any backorder policy. Specify which of these contribute to the quantity exposed for a new order and which are only planning signals. Do not assume that stock at every warehouse is interchangeable: eligibility, location, shipping constraints, and account-specific rules can change what is fulfillable. If a workflow aggregates locations, document the conditions that permit that aggregation and how the chosen fulfillment source is recorded.

Assign ownership to each inventory value and projection

An ERP, warehouse system, OMS, and storefront may all display a number called available, while each represents a different step or policy. Name the authoritative writer for physical counts, allocations, order reservations, expected receipts, and the storefront projection. For each interface, define whether updates are deltas or absolute quantities, which identifier joins the records, what event or version signals a change, how old a value may be before it is flagged, and which team can correct the source. A downstream copy should retain enough context to trace it back to its source and calculation time.

Model reservation changes across the order lifecycle

Treat a reservation as a lifecycle with an owner and evidence: when it is created, adjusted, released, or allowed to expire; which order and line it belongs to; and how partial fulfillment, cancellation, return, or an order edit affects it. Test duplicate and delayed updates so the same order change does not create a second hold or erase a later adjustment. Reconcile remaining reservations against open order quantities and fulfillment state. A cart hold, an OMS allocation, and a warehouse pick are distinct business events unless your documented platform behavior explicitly connects them.

Set a bounded available-to-promise rule

If the business needs a future promise, document the inputs and horizon: current sellable quantity, open demand, eligible scheduled receipts, operating calendars, cutoff times, customer or channel eligibility, and any policy buffer. Decide how the answer changes when a receipt date moves, demand is revised, an order is partially shipped, or the requested location cannot fulfill. Microsoft Dynamics 365 Inventory Visibility documentation describes ATP through configured calculated measures and an on-hand-change schedule; those product settings illustrate why an ATP result depends on the selected measures and configuration. The source is not a universal formula or a delivery commitment for another system.

Use vendor examples to test vocabulary, not to assume equivalence

First-party documentation illustrates materially different inventory models. Shopify defines separate On hand, Available, Committed, Unavailable, and Incoming states; its documentation says incoming quantity is not available to sell until it is received and its state changes. Adobe Commerce describes reservations adjusting stock-level salable quantity as orders are processed, shipped, canceled, or refunded. commercetools documents quantityOnStock and availableQuantity fields and notes that they are eventually consistent with its internal SKU availability used for reservation logic, with a delay of up to 10 seconds. These are scoped examples of those products' documented behavior. They should not be treated as interchangeable field definitions, universal freshness targets, or guarantees about another configuration.

Reconcile differences by item, location, and state

Compare the records that explain a quantity, not only a company-wide total. A useful exception view can compare item and location identifiers, physical counts, open reservations, unfulfilled order lines, held stock, inbound receipts, exposed availability, and each system's update time. Classify a mismatch before repairing it: late event, duplicate order change, missing location mapping, stale receipt date, count adjustment, reservation that did not clear, or an intentional policy difference. Assign the source correction and the downstream replay separately, and keep the before-and-after evidence for the affected records. Avoid blind overwrites that can conceal an unresolved physical or order-state issue.

Test high-consequence order and inventory scenarios

Before widening a release, exercise representative B2B quantities and locations: two buyers competing for the same last units; an order edit or duplicate update; a delayed receipt; an unavailable or safety-stock adjustment; a split shipment; a partial cancellation; a return after shipment; and a location reassignment. Check the expected quantity and order state in every system that participates, then verify alerts, reconciliation, and the operator's repair path. Start with a limited item or channel scope and expand only after owners can explain mismatches and restore the intended state. Set thresholds and rollback conditions from your own operation rather than borrowing a universal number.

Editorial note

This guide is education-only technical research for B2B e-commerce teams. It does not provide legal, tax, accounting, financial, security, performance, pricing, compliance, or individualized implementation advice. Vendor examples describe only the products and behaviors in the linked first-party documentation. Confirm the current documentation, product configuration, and operating requirements that apply to your organization before making a decision.

First-party source register

First-party documentation and standards referenced during editorial review.

Continue reading